Open Source Databases "50% Cheaper"

pete314 writes, "Open source databases can cut the total cost of ownership of a database by up to 60% compared to the cost of running proprietary databases from Oracle, Microsoft or IBM. According to data collected by Forrester Research, the savings average about 50%. Open source databases however still struggle to reach mission-critical enterprise applications because enterprises perceive them to be less secure and stable."

Comments

enterprises also want

[rubycodez]

enterprises also want paralleling clusters and failover clusters. The open source databases are getting there, give it few more years.

Re:enterprises also want

[Anonymous Coward]

enterprises also want all sorts of stupid things. As open source developers also program stupid things, all is well.

Re:enterprises also want

[wiggles]

The one place that the Open Source products have a long way to go is support. Companies don't think mysql and postgres are unreliable, they're just not backed enough. The company I work for could give a rat's hindquarters about TCO -- they just want to outsource their risk so that if something breaks, the CIO/CEO/Chairman has someone to argue with. The Chairman can play golf with Larry Ellison as he tries to get more concessions out of Oracle, but he can't play WOW with the 19 year old kid who added some bit of code to mysql.

And before you say it, MySQL AG is still small potatoes compared to Oracle, Microsoft, or IBM.

Re:enterprises also want

[dingDaShan]

It's true. Any person who is in an IT job does not want to risk going to an open source database (even if it saves the company money) because:
1. He does not want to risk his job if there are problems... if there is an Oracle database, he can just use the Oracle support and tell the boss that they are using the best 2. There is no monetary incentive for the IT professional to switch. If the IT professional would see a benefit to himself that would outweigh the possible problems, there might be a switch, but until then what is the benefit? 3. Some companies just want it to work because they want to do business. Most managers don't want to get involved in database technical issues. They just want results and are willing to pay for it.

Re:enterprises also want

[BSAtHome]

In other words: it is all about appearance and very little about features. The Boss does not make a decisions based on technical merrit, but on charm. The price is a number on next quarter's financial statement.

Re:enterprises also want

[scuba_steve_1]

Ditto this comment. When I rip out a simple system, I definitely lean toward Postgres by default...but when I need a system with any of the following qualities: - clustering / high-availability failover - robust WAN replication - maximized performance - ability to hire experts that can tweak every ounce of performance out of the servers - extensive and complete documentation ...as well as range of other esoteric functions required for enterprise applications, I default to Oracle. Of course, this decision is also almost never an issue since each one of my clients is almost always already in possession of an enterprise-wide license agreement for the full Oracle product suite...because of Oracle's bulldog sales force. The downside to those licenses are, however, that I then have to explain why I would rather use JBoss than OAS. Also, do not downplay the sheer number of people that you can hire that are Oracle DBA and PL/SQL experts. Licenses are cheap compared to the cost of a failed development program...at least in my world.

Re:enterprises also want

[MidnightBrewer]

It's all about the reassurance of having someone definitive to contact for support (or point your finger at) and who has a vested interest in backing their product. People are far more dependable when they're working for money than for charity. Running a business on a product that was developed for free by someone who "felt like it" doesn't encourage feelings of reliability. The same developer might just decide he feels like quitting tomorrow, too.

Re:enterprises also want

[Ewan]

I'd agree with you 100% on your final statement, the $100k cost of Oracle licenses really are irrelevant compared to the cost of a failed project, and indeed the development ecosystems built around Oracle and MS SQL Server can easily save their license costs when compared to the "build-it-yourself" structure of MySQL and Postgres. The various open source databases are excellent, but they often won't save you any more.

Re:enterprises also want

[Anonymous Coward]

In other words: it is all about appearance and very little about features. The Boss does not make a decisions based on technical merrit, but on charm.

It's all about maximizing personal gain for The Boss and The Techs. They aren't spending their own money, and they will get slammed hard if there is a problem that can even be perceived to be a result of them choosing a product that isn't backed by support from a huge vendor. The Boss probably also knows that his employees and other "human resources" are dumb and getting dumber by the day, what with offshoring and the assault on technical workers. He isn't going to risk his ass to potentially save a few hundred thousand dollars per year for the company. Even if it works, it's relatively small potatoes and not worth the risk to him personally.

Re:enterprises also want

[Anonymous Coward]

The Chairman can play golf with Larry Ellison as he tries to get more concessions out of Oracle, but he can't play WOW with the 19 year old kid who added some bit of code to mysql.

Wait a minute. Are you suggesting that perhaps I have the skills that many chairmen lack, which might make me unusually competitive and well-suited for the position?

Re:enterprises also want

[ShieldW0lf]

I would hazard to guess that MySQL AB would rebuild their entire company for you based around a contract 1/10th the amount that say Oracle would charge a large client to participate in their existing generic support infrastructure.

With the kind of dollars big DB companies charge, if you're a big fish and able to solve your business problems with Open Source instead, you could pretty much steer the course of the industry for a fraction of the cost to buy from a big vendor.

With the kind of money being tossed around, you have to wonder why some of these big IT spenders don't just start a subsidary company based around an open source project. It wouldn't have to have making money as it's primary mandate, but rather supporting its owners IT needs first and trying to generate return on investment as an incidental second.

Heh, Postgres is ripe for something like this. Instead of paying a bunch of money to an existing player, if a major corporate IT spender like a bank were to throw down a bunch of money and create a corporate and licensing structure similar to what MySQL AB has done using Postgres' code base, they could probably own their DB vendor for less than it costs to negotiating with the one they have now. The "Who cares if it makes money as long as it saves money" approach.

Re:enterprises also want

[spun]

We have a lot of enthusiasm for open source where I work, and we wanted to move our Family and Child Tracking System off of Sybase and onto Postgress but the major stumbling block is support. We're a state agency and CYA is crucial here. We talked to several companies offering support and they have all sucked so far.

Re:enterprises also want

[Anonymous Coward]

The one place that the Open Source products have a long way to go is support.

True, but slanted. As a programmer at a big company that uses a proprietary database, let me add: the one place that proprietary products have a long way to go is support.

We have lots of experience arguing with database vendors. ("Hey, this call bypasses access control." "Hey, you're right. So be careful with that one." "...") It's not about support. It's about using a big-name proprietary product. That's all: brand. That's a much tougher nut to crack (though as Firefox is showing, not impossible).

Re:enterprises also want

[rhaas]

Just out of curiosity, how did they suck? What are you looking for in a PostgreSQL support company?

Re:enterprises also want

[Anonymous Coward]

Hey!

Its's MySQL AB, not MySQL AG. It's a swedish company and not a german one!

You can buy support 24x7 AND you can play golf (or whatever) with David Axmark, Allan Larsson and Michael "Monty" Widenius. See http://mysql.com/

Re:enterprises also want

[Anonymous Coward]

In the java world you can use C-JDBC which will give you very decent failover, clustering and replication on top of a range of open (and closed) source databases.

BTW ever tried setting up an Oracle 9i RAC?
Try deploying that in an ASP solution to multiple (even small number of) customers and you'll be in a world of shit.

Re:enterprises also want

[MouseR]

Disclaimer: I work for Oracle, though not in any DB department (but rather native Mac applications).

One issue with comparing Oracle with Open Source solutions is the lack of feature parity. And By that, I'm not saying OSS isn't good. PostGres and MySQL are pretty good but they currently lack load-balancing and data replication across multiple data centers. Oracle also has PL/SQLwhich is currently (AFAIK) has no (or short) equivalents in OSS DBs. (PostGRE hasPL/pgSQL but apparently is not feature parity.

When a big corporation already relied on these features, it's hard for them to justify making a move to another DB system. Even another that might have similar features.

Re:enterprises also want

[ThreeE]

People that berate managers for making decisions based on financial statements are naive. Let's see how fast you make a decision when it means a zero vs. non-zero on your paystub...

Re:enterprises also want

[Anonymous Coward]

I would hazard to guess that MySQL AB would rebuild their entire company for you based around a contract 1/10th the amount that say Oracle would charge a large client to participate in their existing generic support infrastructure.

Large clients tend to use the advanced features Oracle has that no OSS database has. I like Postgres (and consider MySQL nothing but a toy) and use it in many places, but I've seen a number of situations where I've told folks they should fork out for Oracle because they really needed it's features. Large companies are going to almost always be the in the latter category. It's the small/medium size companies that can often get by with the basic set of features in an OSS database.

Re:enterprises also want

[jazir1979]

But this really isn't true at all. The companies that have sprung up around open source products only make money by delivering support. Therefore they have to make sure they do a darn good job at it.

To companies delivering software as a product, you are just another customer who can be fobbed off to an offshore call centre and given the run-around for months before finally escalating your problem to the appropriate level.

Re:enterprises also want

[Serveert]

Which features does pgsql lack? I've used plsql+pgsql and for me they're just as powerful. Also for data replication we use Slony(master-slave), with no problems. One feature missing is good partitioning, other than that, postgres has most of the features of oracle. And it's compliant with standard sql unlike mysql. It just requires someone skilled in postgres and there arent many around. Of course with oracle you have all kinds of people to choose from. The problem really isn't features but support and people knowledgable about postgres. How many db admins know how to maintain oracle logs and to point in time recovery, how about for postgres? It's all documented but there are just more oracle admins.

Re:enterprises also want

[kihbord]

I don't think we need to wait that long. Any web site selling something on the Internet is an "enterprise". Try counting e-commerce sites using the Apache, PHP and MySQL combination and you've got a sizable amount of "enterprises" using an open-source database.

Re:enterprises also want

[Ex-MislTech]

Where I worked in a enginering support environment with a oracle front end
we came up with a couple of different new names for oracle.

1. Orsuckle

2....

O=Obtuse
R=RAM
A=Abusing
C=Corrupting
L=Lame
E=Executible

But at the time they didn't want to risk anything else , and so shelled out
$1 million USD a year dutifully, though that was 7 years ago...and I moved on.

Re:enterprises also want

[Anonymous Coward]

As the article stated and previous poster confirmed, the biggest problem is perception. The database will cluster nicely. You just have to convince the CXO that the application will work. I don't remember the kid being 19 years old either, and they audit the code going in (repeatedly), and it isn't unreasonable to expect that in order to do system level work on a database, you should have at least 4 (successful) years of university in Computer Science or Computer Engineering. I don't think they teach it anywhere else. On second thought, they might be 19, but if they already have the 4 years, then they are going onto grad school. Thanks for the disparaging remarks about the developers though!

Re:enterprises also want

[Anonymous Coward]

Oh yes, we all know how well documented slony is.

Re:enterprises also want

[innosent]

As far as I know, the few differences between plsql and pgsql are covered by EnterpriseDB's product (which they provide support for, for a LOT less than Oracle), but other than some minor syntax or naming differences, I can't think of anything you'd actually use that is in plsql that is not in pgsql. Also, PostgreSQL 8.1 has table partitioning. The only major features that are missing are the multi-master replication systems, like RAC. MySQL's cluster is intended to be similar to RAC, but for now is in-memory only, and MySQL is a little new to the large-scale arena and doesn't act quite the same as Oracle and PostgreSQL under load (ACID/MVCC). What we need is RAC for PG, and maybe a little less complaining on forums about COUNT(*) being slow, and more explaining WHY COUNT(*) is slow (MVCC), and why everything else is faster because COUNT(*) is slow (except MIN/MAX, which is now much better for indexed columns). Of course, for the $40,000 per processor you'd spend on Oracle, I'm sure both MySQL AB and EnterpriseDB/PostgreSQL developers would be willing to fix those things. Hell, for a few Oracle processors worth, I'd be willing to add something like RAC to PostgreSQL.

Re:enterprises also want

[DerekLyons]

enterprises also want paralleling clusters and failover clusters. The open source databases are getting there, give it few more years.

The one place that the Open Source products have a long way to go is support. Companies don't think mysql and postgres are unreliable, they're just not backed enough.

Both the parent and the grandparent are correct - but the parent hits the nail on the head with:
 
 

The company I work for could give a rat's hindquarters about TCO -- they just want to outsource their risk so that if something breaks, the CIO/CEO/Chairman has someone to argue with.

*Precisely*. What F/OSS advocates don't get is that TCO is but one factor in determining what package to use.

Re:enterprises also want

[Serveert]

I've never been a fan of multi-master. No matter how fancy they code it, you may still end up with update conflicts. Master - slave is simpler and you know your updates will always work, there are no such conflicts. I use slony to replicate postgres tables via master-slave and it is very fast and efficient. And yes Slony is not well documented. If you code your application around slony then it can work very well.

Last time I looked at postgres partitioning it didn't use a global index like oracle does. On the newsgroups someone asked for the feature and a developer told them basically stop complaining and pony up the money for the feature and he'll code it. So I won't complain. :)

Postgres has been very good to me. I like WAL and how you can do PITR, hot backups, etc. I just think that's cool how you can tar up the database files and it will apply the logs just fine. It's enterprise-ready and free, you can't beat that. I did use mysql for a number of years - that was a stressful time in my life.

Re:enterprises also want

[Anonymous Coward]

Oracle only had full parallel databases from 2 years ago (RAC, and with still not munch install base) What they had until then was more or less a glorified "master-slave" scheme munch like MySQL. So OO DB are not as far as you think from commercial ones.
Enterprises wants to feel confident. Period. The only problem is that with OO DBS they are not confident enough, but that's changing.

Re:enterprises also want

[bigman2003]

You're not using the Enterprise level definition of enterprise.

Re:enterprises also want

[kihbord]

I often wonder what "enterprise level" really mean.

Re:enterprises also want

[plumby]

The one place that the Open Source products have a long way to go is support. Companies don't think mysql and postgres are unreliable, they're just not backed enough. The company I work for could give a rat's hindquarters about TCO -- they just want to outsource their risk so that if something breaks, the CIO/CEO/Chairman has someone to argue with.

That's less about support and more about image.

The quality of the support (such as likelihood of getting someone to be able to fix your DB when it's fallen over at 3am etc) and therefore the level/cost of risk that this incurs is something that should already be right there in your TCO.

What you're talking about is how it will look to your boss (and who you can blame) if things do go wrong. Oracle's support could (theoretically) be 10 times as costly and half as good as some local MySQL support company but if Oracle screw up, it's seen as Oracle's fault not the person that chose them.

Deciding against a supplier because the level of support/liability is not acceptable to you is a good reason to make that choice. Deciding against it simply because you'd look like an idiot if it did go wrong is not a good reason (but it's the one most people seem to take).

Re:enterprises also want

[AliasTheRoot]

It's a nice feature to have, not that it hardly ever works outside of tightly controlled lab situations - but hey.

Re:enterprises also want

[Anonymous Coward]

The ability to do this with MySQL or Postgres is here now. Give it a few years for more or enterprise to realise it.

Re:enterprises also want

[spun]

Well, I'm not part of the DBA team, but I heard they were rude and didn't seem to know much about Postgres internals. We are looking for something similar to the contract we had with Sybase, that's all I know.

Re:enterprises also want

[MouseR]

I'm simply saying that there's no 1 to 1, conversion-free or syntax-change-free way for a big company to move from one DB system to another. Then pile up all the business applications on top of these DBs and you end up with a monster system that requires a LOT of arguments to move from.

Believe me, it's as difficult for Oracle to move IBM DB2 customers to their offering, and for the same reasons.

It's also one reason why there's been a lot of DB/apps acquisitions lately. Oracle's way to gradually move customers to it's offering is to buy competing products, integrate their functionality, cut on overlap. JP-Morgan was on oracle's radar for a while and ultimately, it's PeopleSoft who acquired them. Shortly after, Oracle ate PeopleSoft.

Re:enterprises also want

[eMbry00s]

MySQL AB, [b]A[/b]ktie[b]b[/b]olaget.

Re:enterprises also want

[somersault]

As I'm technically an IT professional then I'd have to say that not all IT professionals think that way (well maybe the ones in large datacentres do, but for small business it doesn't have to be that way). I have sufficient trust in mySQL that I'd use it to power some database apps that I'll likely be writing for different departments to help track and maintain their equipment, that kind of thing. We already have some 'lite' version of Microsoft SQL for an accounts program (one with a rather crappy interface from a company called Access, the product is called Dimensions, it just looks $%!£ and probably performs similarly) we use. We also have a developers version of SQL that the last IT guy got. I'm considering getting MS SQL Server or Oracle, but considering the licensing fees for those, and the fact that if something went wrong with them I'd be dealing with it at ground zero anyway (and the best thing to do would just be to reinstall the product, restore the database etc, or the failure would most likely be completely unrelated to the actual database product...), I don't mind dealing with mySQL, which is fairly proven and works well for small to medium databases. I wrote a small app using PHP a few years ago and ported it to MS SQL, mySQL and also ODBC (yeah I know that just interfaces to other database systems), and found mySQL was the best to work with anyway, though maybe MS SQL has improved since then. mySQL also has the flexibility of being able to be deployed onto a Linux server, which again I'd trust more than any Windows Server, but would require a little effort in setting up proper backups :p

Re:enterprises also want

[CryBaby]

Check out Professional PostgreSQL support

While many of these companies (I'm looking at the North America region) may be too small/unknown for your tastes, companies like EnterpriseDB (and Command Prompt) are well established and have some very large clients including Sony Online and Vonage.

If you're a Solaris shop, don't forget that Sun offers PostgreSQL support, and obviously they can handle support for fairly large corporations.

By the way, here is a picture from the latest PostgreSQL developer's conference. There are very few "19 year old kids" in that group.

Re:enterprises also want

[CryBaby]

Talk to Sun if you're on Solaris, EnterpriseDB if you're on Windows or Command Prompt for Linux/BSD.

Re:enterprises also want

[spun]

Command Prompt was the one we talked with and didn't get a competent vibe from.

Re:enterprises also want

[Anonymous Coward]

"It's all about the reassurance of having someone definitive to contact for support"

Bullshit. If so it were, using MySQL would be as easy as geting a "gold support contract" from... surprise! MySQL AB an incorporate enterprise that doesn't contract hippy-like pimply teenagers.

It's, as always, about perception.

Re:enterprises also want

[CryBaby]

Command Prompt has several PostgreSQL developers on staff so, as a company, I think they know plenty about PostgreSQL internals and are probably highly competent. However, I can't speak for their support offerings and I doubt that the developers I'm talking about work in the support department. Sorry to hear they didn't live up to your expectations.

I get the feeling that EnterpriseDB understands a lot more than Command Prompt about what companies need in terms of support and what it takes to get them to switch from Oracle to PostgreSQL. Whether or not they are competently addressing those needs is another matter (I don't have any personal experience with them), but I would say it's worth a phone call if your company is serious about evaluating PostgreSQL.

Re:enterprises also want

[turbidostato]

"PostGres and MySQL are pretty good but they currently lack load-balancing and data replication across multiple data centers."

Very good Mr Marketroid, but you didn't even read the post you are answering to.

Just to save you the work, the nut of it was:

You Can Program Those Lacking Features For A Fraction Of The Cost Of Going Oracle.

Clear enough?

Re:enterprises also want

[turbidostato]

"I'm simply saying that there's no 1 to 1, conversion-free or syntax-change-free way"

And then, the article is "simply saying" that while it might be no 1 to 1 conversion-free way, there is a 1 to 0.5 money free one (of course *when* such a move away is doable... warm regards to closed software lock-in strategies).

Re:enterprises also want

[spun]

Thanks for the info. I'm a little out of the loop on this, being the Linux/BladeCenter/VMWare admin, but I talked to our DBA group and their manager said he will be meeting with EnterpriseDB on Monday. And Command Prompt isn't out of the running yet, from what I hear it was just one guy we talked to over there was surly or something. We are quite serious about moving to open source in general, and particularly for our database. We are a state agency and we have a very nice case management system that we are trying to move off of HP/Sybase to Linux/Postgres. Hopefully, then we will open source the whole application.

Re:enterprises also want

[innosent]

Multi-master is certainly not without problems, as it inevitably involves some compromise at the application level to at least minimize the potential for conflicts (which is why I handle async replication outside the database, with each master assigned their own ranges for things like auto-increment columns [and reduce the number of auto-increment columns as much as possible], and triggers to track changes that need to be synchronized), but some applications cannot do without it. Things like banks and hospital groups, or basically anything with multiple mission-critical sites that need access to all data instantly, must have updateable databases onsite, just in case the network connection to the primary is lost.

Re:enterprises also want

[kabz]

I work with both Oracle 9.2 and SQL Server 2000 on a daily basis, and the company I works for doesn't write much code that gets out of the compatibility zone that these two DBs. Of course, the solid reasoning behind this is that consulting companies produce software and we do not want to have the expense of supporting two very different code bases, and we cannot afford to say no to clients who want to run either Oracle or SQL Server.

Companies like this will eventually start supporting the open source databases, but probably only as customers demand. Currently this isn't much demand as our clients are pretty much tied into one of the big two/three, but it will surely come as big companies realize that if Google depends on MySQL, then maybe they could too.

0% savings for me

[Crudely_Indecent]

For those of us who can't afford to run a commercial database package, and have been running open source databases from the beginning, this isn't news. MySQL and Postgres are your friends.

Re:0% savings for me

[Anonymous Coward]

Free, eh?

As Bruce Perens -- famed open source advocate recently turned real world cynic -- once said: Linux is only free if your time has no value. Or, in this case, Open Source Databases are only free if you don't value your time. Your TCO is NOT zero, regardless of what you "paid" for the software.

Re:0% savings for me

[Kenja]

"MySQL and Postgres are your friends."

No, they're my bitter enemies whom I cant yet kill because they still serve a purpose.

But one day... their uppance will come!

Re:0% savings for me

[leamanc]

Which brings up a good point. How many "enterprises" need an Oracle or DB2? MySQL and Postgre, despite their obscure limitations that really only matter to ubergeeks, can work just fine for non-Fortune 500 companies. Heck, they would (and do) work fine for some of the big companies. Most of the small businesses have limited in-house IT, but usually have a guy or two that can learn PHP and tie into an open source SQL with that.

The big boys are the only ones who need the big DB vendors, and even in that case, it's more so they can write off the cost of purchasing the licensing and paying the implementation team. Better to give the money to other big businesses than to Uncle Sam. (Obviously written from a USA-based perspective.)

Re:0% savings for me

[Crudely_Indecent]

only free if you don't value your time

I do value my time. Unfortunately, I do not have a large bank account to draw from. I can afford to invest time into my company, while I can't afford to invest the thousands necessary to support an Oracle infrastructure. Time I have.

Re:0% savings for me

[GuidoW]

I don't think this holds true for database systems.

As far as I can see, the big databases like Oracle or Sybase are targeted at big corporations that don't have any second thoughts about hiring a full-time DBA - and as a consequence you will need one to use these products. Okay, maybe not a fulltime DBA, but at least a trained professional (trained for that particular application) to invest quite some time to even get stuff going.

For Postgres, OTOH, you'll just need someone smart with general knowledge of SQL and of the platform it's going to be running on - he needn't even have used Postgres before.

Re:0% savings for me

[glwtta]

Open Source Databases are only free if you don't value your time.

Dude, have you actually used both Postgres and Oracle? There is not an entity on this earth less respectful of my time than Oracle (well, maybe ClearCase) - the thing is an absolute nightmare to administer. Sure, it needs the complexity because of its advanced scaling capabilities; but most of us are not amazon.com, and never will be.

On the other hand, the administrative overhead of running Postgres is damn near 0% (MySQL is a different story entirely of course).

Sure we are a small company, and only have under a TB of data in our databases, but there are a lot of companies in the same position who shell out ridiculous amounts of money for Oracle (only for the name-brand, nothing else), and then someone ends up stabbing themselves in the eye in frustration (might be a slight exaggeration). Or else pay for a full time DBA; I've worked for a company with 3 developers and 1 full time Oracle DBA - that's just nuts.

Re:0% savings for me

[wetfeetl33t]

Open Source Databases are only free if you don't value your time.

I dunno about that. A few years back we started running MySQL. Took a few hours to get set up and running, and it still needs to be maintained and upgraded, but systems like MySQL and PostgreSQL just seeem to run and run and run without any intervention... I've almost forgotten the root password because I never have to go in it.

Re:0% savings for me

[Anonymous Coward]

I've worked for a company with 3 developers and 1 full time Oracle DBA - that's just nuts.

Considering most developers I've seen know nothing, or nearly nothing, about data management and can only write basic, entry-level SQL, a full-cycle DBA is a very good idea even with only 3 developers. Of course most object-oriented religious zealots will now argue that relational technology is dead so what's the deal...

Re:0% savings for me

[Vancorps]

What kind of administrative overhead does Oracle require for you? I set it up, takes about 20 minutes. Apply my scripts to size the server resources and I'm away. Sometimes it'll take about an hour to copy a TB of data but that's pretty damned fast. Once that is done the server is up and running and I never need to touch it. The only reason I even ever login to my Oracle boxes is to check and make sure backups are running properly and to ensure there are no sudden changes in disk space usage. Haven't had any so far but it's always good to check.

Of course now I monitoring software in place so I don't even need to login to the box to know it's okay.

Not going to say it's no overhead, if you run into a quirk of the system is can take a little while to solve the problem. It is usually a case of having to think like an Oracle developer.

It's been a few years since I've played with Postgre and mySQL so I won't really comment on the administrative overhead they require. I had a terrible first experience with mySQL and an even worse second experience. This time around I thought I'd give it one more shot since it finally support multi-master replication but when it came to having to support 20 servers and every workstation in the company by myself I start to look at Oracle support and their metabase and I feel better about the decision we went it. Oracle streams became the defining reason we chose since it provided a very convenient method for us to offload data to various parties that need access to a subset of data. These kinds of features are sorely lacking in the OSS DBs I've seen. Although that could have changed. I last evaluated over a year ago. Not sure I would rely on something that new though even if it were released since then.

I might add that Oracle for us was at least cheaper than MS SQL and provided better/more refined features. Administrative overhead is incredibly low to the point where my database is the least of my worries and it is indeed mission critical.

Re:0% savings for me

[Anonymous Coward]

Some clarifications about Oracle.

[1] Has been fairly easy to install since version 9.
[2] There is a free version of Oracle which is extremely simple to install on linux or windows and works out of the box w/out any DBA. (Limitations - 1 CPU, 1 Gig memory, 4 Gig user data ... fine for many small apps.) Includes backup/restore scripts.
[3] "Standard One" edition (which is fine for many applications) costs 5K per processor, not >100K.
[4] Comes with a great 2-tier web development framework installed by default. (Formerly called htmldb, now Oracle application express.)
[5] Self-tuning and smart defaults has diminished need for DBA on smaller apps.
[6] PL/SQL is a very strong procedural language for logic close to the data and for data processing in general (e.g., exception handling, pipelined table functions, external tables, merge cmd, exceptions into clause, sql xml functions, sql analytic (e.g., rank(), cust_dist() ) functions, etc.) You can implement complete systems with only the database platform.
[7] Oh, should I bother mentioning multi-versioning concurrency control (non-blocking reads) or flashback query/table/database?
[8] Strong pl/sql api's (e.g., job scheduler, compression, http, email, encryption, debugging, ldap, xmltype, file, etc.)

I don't work for Oracle. I am just a developer whose used it for a number of years. It is an incredible piece of software engineering with over 30 years of maturity. The SQL optimizer is amazing piece of computer science. Open source databases are real and may catch up some day. But I prefer to work with the best.

And the truth is: it isn't necessarily expensive or hard to use.

Re:0% savings for me

[0racle]

For Postgres, OTOH, you'll just need someone smart with general knowledge of SQL and of the platform it's going to be running on - he needn't even have used Postgres before.

That is a recipe for a poor performing database and a disaster. It doesn't matter what RDBMS you use, a database is a complex object and not something you can throw together and expect it to work right, if at all.

Re:0% savings for me

[CastrTroy]

This is where you will really see the difference. There are some people who have a valid need for Oracle. But there are also a lot of people who are paying for Oracle who could do the same thing with MySQL or PostGres. There's a lot of MS SQL Server licenses that are only being paid for because the only alternative they can think of is Access, which isn't much of a database. They don't really need SQL Server (and definitely not Oracle), but they need something more than Access. This is the market that PostGres and MySQL should be aiming for. The other issue is that they don't want to have to support more than 1 database. If they need Oracle for 1 project, they may choose it for all, simply because that way, there's only 1 DB to support.

Re:0% savings for me

[jbplou]

I might add that Oracle for us was at least cheaper than MS SQL and provided better/more refined features.

How could this be? SQL Server is half the price per processor and provides most features out of the box, Oracle requires add ons. I'm not going to say SQL Server is better but it does provide more bang for the buck.

Re:0% savings for me

[tcopeland]

> For those of us who can't afford to run a commercial database package,
> and have been running open source databases from the beginning,
> this isn't news. MySQL and Postgres are your friends.

Right on. Those of us with over 16 million records in a PostgreSQL database are pretty happy with what we're seeing. And we're willing to run the same database on production servers as well.

Re:0% savings for me

[Anonymous Coward]

Until not that long ago, MySQL thought February 31st was a valid date.

Installing a difficult to use app is a pain. Data corruption is absolute hell.

Re:0% savings for me

[Vancorps]

Per Processor for the enterprise edition of SQL 2005 vs what you get with the Standard edition from Oracle which didn't require us to go per processor. With Oracle we could have a small number of named users and have access to all the 64bit addressing and processors we can shake a stick at. Plus we can cluster them which was the big fault for MS SQL. We caught Oracle at a good time, they came down a lot over the initial price quote they gave us.

The only expensive part was paying for access to the metabase. Part of me thinks I could get by with just my experts-exchange account but that support is really nice when you need it.

Re:0% savings for me

[Mongoose Disciple]

Sure we are a small company, and only have under a TB of data in our databases, but there are a lot of companies in the same position who shell out ridiculous amounts of money for Oracle (only for the name-brand, nothing else), and then someone ends up stabbing themselves in the eye in frustration (might be a slight exaggeration). Or else pay for a full time DBA; I've worked for a company with 3 developers and 1 full time Oracle DBA - that's just nuts.

My consulting experience holds this to be true as well. I've worked on a number of projects over the years that involved an Oracle database; exactly one really needed Oracle for what they were doing. Everyone else essentially bought a Formula 1 racer complete with pit crew to drop their kid off at soccer games. Most of those projects would've been fine with mySQL. (And, I grant you, for that one massive enterprise project that really needed Oracle and had a sizeable team of DBAs to tune and support it... it worked great and I can't imagine having used anything else in its place.)

As a developer, I prefer to work with MS SQL Server given a choice, just because, specifically from the perspective of a developer who needs to work with the DB as a developer does, but who will probably never need to administrate the DB, which are the case most of the time for me, I find it the easiest to work with. I spend more time working on the parts of the problem that are interesting to me and less time massaging a date into a format that Oracle likes. (Or similar details for Sybase or mySQL... I admit I've never used postgres)

Re:0% savings for me

[kimvette]

Funny, I find KDE far more productive than Windows' explorer desktop.

I find SQL Server to be far more productive than Oracle, because in addition to the (weak) CLI and ODBC, SQL Server has a usable GUI.

I find MySQL far more productive than either because in addition to the CLI, there are FULL GUI apps, FULL web front ends, AND ODBC. Yeah, there's a web front end, but it's kind of weak.

The catch with the open source solution is: you have to know that mysqlcc, mysql-administrator, mysql-query-browser, and phpMyAdmin exist, and you have to be inclined to RTFM because you probably won't stumble across them by randomly poking around the Start menu until you find it.

Re:0% savings for me

[Anonymous Coward]

Once that is done the server is up and running and I never need to touch it.

never add tables or change columns as requirements change? Did you know in postgres you can do that in a transaction. You can drop whole schemas, even entire databases in a transaction, and then roll it back if you need to.

Re:0% savings for me

[clambake]

No, they're my bitter enemies whom I cant yet kill because they still serve a purpose.

But one day... their uppance will come!

Will this be when postgres puts out 8.2.final with built-in repliaction? Oh wait, you DON'T want to get rid of oracle?

Re:0% savings for me

[bigman2003]

I hope that your company is database related.

Because if you own *any* other type of company, and you are wasting your time saving money by using a free database, (And from your message I infer that you are spending more time) then your company will never be successful.

As a person who has tasted both success and failure in business, I can tell you that running a business is not the same as having a hobby. If a business owner is doing things like screwing around with a database to save money, then you may not have a viable business.

Spend your time building your business.

Re:0% savings for me

[dwandy]

I spend more time working on the parts of the problem that are interesting to me and less time massaging a date into a format that Oracle likes.

non-strictness is a MS problem, not a feature.

The last thing I want is for the database to 'guess' at what I meant, like Word does when 'guessing' how I want my document formatted, and stubbornly refusing to let me correct it :(

Allowing software to 'guess' when you should be specifiying is a quick way to introduce fun and unpredictable errors into your system.

Re:0% savings for me

[LDoggg_]

The catch with the open source solution is: you have to know that mysqlcc, mysql-administrator, mysql-query-browser, and phpMyAdmin exist,

How is that different than oracle? Oracle provides several different front ends that you don't seem to know about.

That said, I love mysql. I just wish that the developers would add sequences. Sequences are not a very complex construct, but they are extremely useful. Oracle and postgres have them (I can't speak for sql server as I have no interest in ever running windows servers)

Re:0% savings for me

[AliasTheRoot]

expert-sexchange can rot in hell, had enough of googling stuff and being linked to some webpage that will answer my question for a low low subscription rate.

Re:0% savings for me

[Mongoose Disciple]

How much time have you spent working with SQL Server?

I haven't ever had a problem of this type with it. You can specify what you need to. It's just less frustrating to work with as a developer for a number of reasons.

(I also wrestle with Word's auto-formatting. This isn't anything like that.)

Re:0% savings for me

[kimvette]

I'm aware of them, it's that Oracle's GUI sucks compared to the other players' options. Or, at least they did the last time I used Oracle (2001).

Re:0% savings for me

[Anonymous Coward]

[7] Oh, should I bother mentioning multi-versioning concurrency control (non-blocking reads) or flashback query/table/database?

You should mention it, as transactions and locking are the primary reason many of us use a dbms. But Postgresql's locking and version control are very similar to Oracles's from a functionality point of view-- they even have some advantages. So far as I know Postgresql does not support the flashback feature.

Oracle has gotten better than a few years ago--- when ever configuration parameter defaulted to something insane--- but IMO Postgresql is still easier install well and run well.

Re:0% savings for me

[jedidiah]

This is just stupid. If you have no interest in
performance, and just want something you can just
throw somewhere and forget about then Oracle will more
than adequately accomodate you. The relevant documentation
is quite readily available and you will find that the
combination of the complexity of Oracle coupled with the
ease of finding suitable documentation will make it much
easier to deal with than Postgres.

            You also won't have to deal with Postres's assinine
internal procedural language.

            The DB guru is not for the day in and day out crap
he's for when the shit hits the fan and you don't want to
be dependent on someone outside the company with no
particular loyalty to you.

              Do you skimp on qualified sysadmins too?

              If the facility brings value to the company
then any sane management will see to it's protection.

              Any idiot can use the Oracle installer to
create a database and start generating Cartesian
joins. In this respect it's no different from postgres.

Re:0% savings for me

[jedidiah]

Such (schema) changes are no more difficult
under Oracle than they would be in mysql or
postgres. The syntax may even end up being identical.

        Now if you want to roll back an Oracle database to any
previous point in time or change number that's no problem
either. Although ideally you would avoid the need for this
sort of thing with proper change control.

Re:0% savings for me

[jedidiah]

If you really need the GUI then...

        You shouldn't even be a DBA.
        You shouldn't be a Sysadmin either.
        And you certainly shouldn't be a developer.

        I can certainly understand the interest in
not being "inconvenienced" but if someone is
bright enough that they have any business coding,
then managing a simple dev database should not be
an issue.

        This isn't exactly a web browser or word
processor we're talking about here. If you can't
be bothered to pick up the basic DBA skills than
you probably can't be bothered to write good
procedural code or SQL that won't cripple the
database.

Re:0% savings for me

[jedidiah]

If you are willing to commit your tech management
& staff to be at Oracle's disposal 24/7 until a problem
is solved, then the EE version of Oracle makes sense for you.

            That is the level of commitment you are expected to
make when you open a level 1 support ticket with Oracle.

            OTOH, you could just use the SE version for cheap.

            You don't have to pay 500K for the stuff.

Re:0% savings for me

[jedidiah]

16M records isn't squat. I've had Oracle dev databases with 500M.

Re:0% savings for me

[ckaminski]

No, in general, a database is EXACTLY something I should expect to be able to throw together and have "just work". The fact that I may be able to tweak it to death and/or improve queries to improve my query throughput is a bonus. But if I start shoving SQL into a SQL database, I'm going to get SQL out, and that's how it should be.

Re:0% savings for me

[somersault]

I've written a couple of small database apps (task management, timesheet entry and reporting), and have managed to setup databases fine for those purposes. I admittedly did take a course on databases at uni, but it's not that hard. For large enterprise databases you'd want experts, but when it comes to smaller applications, then all it takes is a bit of research and planning, and you can have things working fine. I agree that knowing how to do SQL queries and knowing your way vaguely around Linux or Windows isn't the best starting point, but saying it's always going to be a disaster every time is a bit much - your average Computer Science student should be able to get a small database application up and running in a couple of weeks even with no previous experience of the specific system, and once they've learned their lessons for that, they're ready to move onto larger systems. My main concern when writing my first app wasn't performance anyway, it was having adequate security.

Re:0% savings for me

[Anonymous Coward]

we're running postgres 8.1.4 with a table that currently contains ~1.3 billion rows, about ~300 bytes per row , using partitioning. i do hope they soon automate the partitioning setup. we have a 4 person dev team and i don't think we could use oracle without hiring a full-time dba, tried it before... i'm actually quite surprised how well postgres has been running maintenance free for us on a such a large db. some of the queries could be planned better.

DUH!

[zappepcs]

Of course its that much cheaper... license fees alone are more than half of maintaining a reasonable Oracle installation... this is news? how?

Re:DUH!

[j-pimp]

Of course its that much cheaper... license fees alone are more than half of maintaining a reasonable Oracle installation... this is news? how?

Does that mean the licensing fees are to expensive or the maintenance requirements are minimal?

Re:DUH!

[Anonymous Coward]

big news dpt
Open Source to be cheaper than M$

Re:DUH!

[RAMMS EIN]

The news is that someone has quantified how much cheaper the open source solutions are. Obviously, a free database costs 0% of a commercial database when you just look at buying/licensing the software, but that's not the only cost you have. This study factors in other costs of running a database, and then concludes that "open source databases" are 50% "cheaper". (Scare quotes, because I haven't RTFA, so I don't know what they are comparing with what)

SQL Server = Almost Free

[blaster151]

I just don't get it. TCO and tool support are tightly linked. Most open source database products, including MySQL, seem to require quite a bit of digging and cobbling together to set up and maintain. Microsoft SQL Server has fantastic tool support, no command line experimentation required. An experienced DBA can set up a new installation in a couple of minutes. And there's even a free Express Edition available for entry-level dabbling. The cost of a database license is pretty minimal over the long haul (referring to SQL Server, not the abominable Oracle). The real cost is in the time spent compensating for whatever your database platform's tool support shortcomings are. I love Microsoft SQL Server for this reason: I rarely have to reinvent anything.

Re:SQL Server = Almost Free

[sstern]

An experienced DBA can set up a new installation in a couple of minutes.

An experienced DBA can set up MySQL with many useful tools in a matter of minutes, too. And you can pay him more because you're not paying Microsoft.

The important question is whether you've bought an application that requires a specific database. As I look at various enterprise apps, they don't come stand-alone, but come in versions tailored for specific databases. If more people said "Do you support MySQL?" we'd see greater use. Chickens and eggs, but you have to start somewhere.

Re:SQL Server = Almost Free

[profplump]

First, if you're afraid of command-line work, you aren't running Oracle, or anything else in that class.

Second, unless you're doing something out of the ordinary, simply installing mysql or postgres in the same way you usually install programs (be that apt-get, rpm, MS Installer, etc). is all you need to get the database up and running. The same is true of the GUI tools to manage the database -- the Windows installer for postgres includes PgAdminIII in the same package as the database itself.

I'm not bashing MS SQL Server, but let's not pretend it has some magical ease-of-use that doesn't exist in other packages.

Re:SQL Server = Almost Free

[LWATCDR]

"Most open source database products, including MySQL, seem to require quite a bit of digging and cobbling together to set up and maintain."
What are you talking about. MySQL can be installed by a monkey and is well documented. Often you don't even have to do anything to install it but check a box on your Linux installer.
If you must have a GUI here http://dev.mysql.com/downloads/other/mysqlgui/ is one of many.
I am glad that you like SQL Server. If it fits your needs more power to you. However MySQL really isn't hard to use or setup. Postgres is a little more difficult but I actually prefer it for most of my projects.
Ease if install shouldn't really matter a lot for a Database server. Hopefully it isn't something you do on a daily basis.

Re:SQL Server = Almost Free

[kpharmer]

I actually found sql server to be quite expensive - from licensing (which was running > $80k for a 4-way on enterprise edition) to labor.

The lack of command-line features meant that many operational activities that could be automated required a dba to manually do the job via the gui. And lets not even talk about how you had to completely recreate DTS packages when promoting them from dev to test to prod...

So, there are labor savings that you can get on sql server vs oracle, db2, postgresql, etc - but the lack of a command line interface wasn't a driver in my experience.

Re:SQL Server = Almost Free

[Petaris]

Setting up MySQL and Postgres is very simple on Linux distros and even Windows. On Linux it requires just one command to install (eg, apt-get install mysql-server, emerge mysql, etc), on windows, its download, run setup.exe, done. And there are tools out there that make managing them simple for those who don't know how to use a command line (eg, phpmyadmin, or for windows mysql's own manager app). I am not an expert db admin and yet I have three mySQL installs and one Postgres install that I maintain, easily. As for an experianced DBA, he or she should be capable of setting any of the options up easily and should know how to use the command line managment of the db they work with.

Re:SQL Server = Almost Free

[chochos]

PostgreSQL has always been really easy for me to install. On Fedora, yum install postgresql-server... on Debian there are packages for postgresql server; on Windows, PostgreSQL 8.0 has an install wizard.
Maybe it's more difficult to set up if you don't like to edit the pg_hba.conf and postgresql.conf files, but other than that...

Re:SQL Server = Almost Free

[blaster151]

Actually, we are running Oracle--we're a software shop and we have to support both MS-SQL and Oracle installations. Oracle just sucks, in the opinions of myself and my colleagues. Sucks in terms of the amount of workarounds to do things that MS-SQL just does automatically. I'm not afraid of command-line work, but I am afraid of wasting time. That spells death for our company and threatens what we're trying to accomplish. After years of MS-SQL use, and about one year or so of trying to mirror the same tasks on Oracle, I've come to hate the product . . .

Re:SQL Server = Almost Free

[xQx]

Funny you should say that... We use Microsoft products (ie. windows servers, MS SQL, Exchange etc.) over the open source or *NIX products because in country areas it's *MUCH* easier to find staff who understand Microsoft products.

In small businesses when you have one person who is your network admin / systems admin / dba / tech support it is important to have products that are easy for a (skilled) tech to figure out.

Sure, a MySQL expert can script and use the Open Source tools faster than a Microsoft Expert can Microsoft tools... but what's the learning curve on MySQL's CLI, or X-Windows tool for a systems admin (ie. not dba) compared to Microsoft's SQL Enterprise Manager?

I'm sorry to say it -- it may be slower, but every time I want to create a database I'm going to be wanting to right click on something and select the 'create database' or 'attach database' option... not try to remember if the command is 'create database [databasename] [physcial file] [log file]' or 'create database name=[databasename] file=[physical file] log=[logfile] set recovery=full'

So how does the TCO line up when you compare what your existing tech can do with Microsoft SQL, vs. paying $250 an hour for the _ONLY_ contractors in town who understand *nix or trying to bribe a DBA from the city to move to the country and work for you?

Re:SQL Server = Almost Free

[profplump]

Well then I'll revise that to clarify my point:

command line != wasting time

But as an Oracle user myself, I agree, they make things hard, particularly in the common case where you don't care about the 10,000 different ways to sort dates or whatever mundane setting has bad or no defaults.

Re:SQL Server = Almost Free

[marcello_dl]

> Most open source database products, including MySQL, seem to require quite a bit of digging and cobbling together to set up and maintain. Not on linux IMO. Of course if the windows port of postgres used to require, IIRC, cygwin it's not really a db fault is it? Never had problems with a couple of local and remote installation of Postgres. And the fcgi app connecting to it survives live updates of the OS and the db code. >Microsoft SQL Server has fantastic tool support, no command line experimentation required. Well I won't ever doubt it. My support tool is a mailing list which is not fine for enterprise. A command line tool is very welcome, for example when depl oying to a shared server which needs the ram for the daemons, not for pretty GUI apps. Also, autocompletion and help of psql seems adequate to me. I often prefer it to firing up the GUI tool. > An experienced DBA can set up a new installation in a couple of minutes. Under linux the package system usually sets up a new installation. One just has to put something in the ACL of the db to allow the app an appropriately secure connection if needed. An experienced DBA sets it up in seconds, me in minutes :)

Re:SQL Server = Almost Free

[glwtta]

Most open source database products, including MySQL, seem to require quite a bit of digging and cobbling together to set up and maintain.

Actually, that's just MySQL, the others are comparable to SQL Server as far as the complexity of a "basic" setup goes. Of course one could say that the time needed for a new installation is just not important in the long run.

I've heard mostly good things about the newer SQL Server versions from people, unfortunately you can't use it alongside a free/open application stack, unless you want to support two entirely different platforms. Which is a bit sub-optimal for the smaller shops that these things are targeted at. That's the main problem with Microsoft, it's all or nothing with them - does not play well with others. Oh, and of course for complex queries Postgres kicks SQL Servers ass, performance-wise.

Incidentally, an experienced DBA loves the command line more than his children, and would never be so derogatory as to call it "experimentation" :)

Re:SQL Server = Almost Free

[Anonymous Coward]

Which version are you using?

SQL Server 2005 is MUCH cheaper than $80k for a 4 way server, and there is nothing that cannot be scripted via TSQL and run via the command line.

Query Analyser is your friend.

Re:SQL Server = Almost Free

[jacksonj04]

I see you've never worked with big databases. Clustering? Redo logs?

Re:SQL Server = Almost Free

[Nkwe]

The lack of command-line features meant that many operational activities that could be automated required a dba to manually do the job via the gui.

Um what do you mean about the lack of command-line features? SQL Server has only one interface and that interface is SQL text sent to it from a client. The only thing that all the GUI tools do is write SQL statements for you and send them to the SQL engine. Anything that the GUI tools can do, you can do as well from the command line (ISQL / OSQL / Query Analyzer / Any software you write that can issue SQL requests.)

In fact part of the very definition of a relational database requires that all communication with that server is via the standard query interface and that there are no special "tools" that directly access the underlying data store and bypass they query engine.

Re:SQL Server = Almost Free

[Itninja]

I'm sorry to say it -- it may be slower, but every time I want to create a database I'm going to be wanting to right click on something and select the 'create database' or 'attach database' option... not try to remember if the command is 'create database [databasename] [physcial file] [log file]' or 'create database name=[databasename] file=[physical file] log=[logfile] set recovery=full'

Let me introduce you to a friend of mine: phpmyadmin (cost: $0)
Then you even be sitting on your tractor out in the cornfield and still use your laptop to create and administer the db. And if you don't like the few extra seconds it take for the screen to refresh, use phpflashmyadmin (cost:$ 0)....

Re:SQL Server = Almost Free

[ednopantz]

yeah but why let ignorance get in the way of MS bashing?

Re:SQL Server = Almost Free

[Achromatic1978]

use phpflashmyadmin (cost:$ 0)....

Not to be an ass, but phpFlashMyAdmin is not $0, but $5.

Re:SQL Server = Almost Free

[Cico71]

The lack of command-line features meant that many operational activities that could be automated required a dba to manually do the job via the gui. And lets not even talk about how you had to completely recreate DTS packages when promoting them from dev to test to prod...

Sorry dude, but you are just making up arguments or you never even bothered to open the documentation.

Every single task you can do via the GUI is done behind the covers either by issuing Transact-SQL commands, or by executing stored procedures. Of course all these commands can be issued from one of the command line interactive tools (isql.exe for versions before 7.0, osql.exe for 7.0 and 2000 or sqlcmd.exe for 2005).

You can also use profiler (via the GUI, or server-side tracing with the SQL trace infrastructure) to capture exactly which commands are issued by the GUIs. In SQL Server 2005 a lot of GUIs behave like a flight recorder and have a button to generate the Transact-SQL code to execute all the commands for a given task.

Then of course you can use any scripting language that supports ODBC/OLE-DB/db-library (to a certain extent) to issue the very same commands. Speaking of scripting languages, there's also a COM object model (SQL-DMO) and a .NET extension (SMO) to interact with a SQL Server instance.

In the box there's also SQL Server Agent a scheduler (and event listener) that can be used to execute all these scripts you can create.

Just a small set of configuration options is not available through Transact-SQL or stored prcoedures (eg. service and net library configuration) but you can still do it via scripting (WMI, DMO, SMO etc.)

About DTS packages what can I say... if you recreated then to move them from test to production then you should have spent your time reading a little bit more the documentation. DTS Packages can just be saved to any other server, file of even VB source. DTS has a rich object model and even if you had to just change connection items you could have written a small script to do it programmatically (but there are also other methods)

Re:SQL Server = Almost Free

[Anonymous Coward]

i'd agree sql server was almost free, but i've spent 3 weeks trying to get it set up on linux and it just won't do it.

what am i doing wrong?

i know you'd include msft lock in as a cost of sql server if it only ran on windows, right?

Re:SQL Server = Almost Free

[kpharmer]

> Which version are you using?

SQL Server 7 & 2000

> SQL Server 2005 is MUCH cheaper than $80k for a 4 way server,

No, that depends on which version you need to use: http://www.microsoft.com/sql/howtobuy/default.mspx

So, if you've got an internet-accessible search engine running sql server (wasn't my decision) then you can easily blow over $80k to license a four-way server. In fact the original estimates we got were over $100k.

Re:SQL Server = Almost Free

[quantum bit]

Maybe it's more difficult to set up if you don't like to edit the pg_hba.conf and postgresql.conf files, but other than that...

Editing pg_hba.conf is nothing compared to having to configure a TNS Listener instance and names files on all the clients, even with (or perhaps in spite of) the "help" of the wizard.

PostgreSQL is an absolute breeze to install and configure.

Re:SQL Server = Almost Free

[Anonymous Coward]

Parent is right, SQL2005 standard edition drives up to 4 CPUs and supports failover clustering. Unless your users authenticate against the database (for example, you require registration and then login from unauthenticated web users), you'd be fine with SQL2005 standard.
Too little, too late, I know. However, most acquire SQL proc licenses in some volume licensing scheme (the co I work for, included), from the OEM, and hence you don't spend THAT much.

Which doesn't change the fact that CPU licenses are bloody expensive, from a /.-ter's pow. Not so much from a co's, where it's THE life and the value of the company.

Re:SQL Server = Almost Free

[roster238]

MS still does not charge by core but rather physical CPU only. Based on that you can run a 4-way quad core server for the same money which makes MSSQL compare quite favorably to Oracle or DB2. I have DB2, MSSQL, MYSQL, and Progress installed in my company and MSSQL is the only one that does not require daily attention. All of the other databases have dedicated DBA's while MSSQL is maintained by the system admins when anything is needed (rarely). It runs more databases (mission critical) than anything but DB2 and the billing software is on MSSQL because it has to work. MYSQL has three databases and has issues frequently. Most folks tell me that if MYSQL has issues it's because my DBA isn't very good. I agree but the guy who works on MSSQL isn't the sharpest tool in the shed either and it always works. When it come to TCO MSSQL wins hands down.

Re:SQL Server = Almost Free

[quantum bit]

I see you've never worked with big databases. Clustering? Redo logs?

With MySQL, forget it.

PostgreSQL has "redo logs" (it calls them Write Ahead Logging). You can do log archiving as a form of incremental backups, or you can copy the WAL segments to another server for hot standby -- we have an HA cluster at work that uses this method; more reliable than a SAN-based cluster as there's no single point of failure. You can do point-in-time recovery and switch between different timelines if necessary.

There's also asynchronous replication via slony (good for on-line read-only cache servers), or synchronous replication with several other methods.

Oracle has an advantage in the multimaster clustering arena. It's not until your databases get really big that you need that, but the that can be a very tricky subject as most databases aren't designed with the idea of not having everything always consistent. PostgreSQL can't do that yet but is actually gaining ground here -- you could in theory do something similar with two-phase commit over separate servers.

Last I checked Oracle was still slightly better with GIS and spatial data. PostGIS has historically had some indexing problems, but I admit it's been a while since I've looked at it.

One thing you can't yet get with PostgreSQL is the privilege of paying oodles of money to Larry & co for unhelpful support staff that usually have trouble understanding what the problem is :-P

Re:SQL Server = Almost Free

[quantum bit]

Oh, and of course for complex queries Postgres kicks SQL Servers ass, performance-wise.

Has anyone done any tests lately to see if that is still the case? EULA be damned, post as an AC if you have to :P

I hear SQL Server 2005 finally got MVCC and did away with row locking. Hard to believe even 2003 still locked rows for every write, but there it is. Even without that edge I'd like to believe PostgreSQL will still kick ass on complex queries, but haven't done any empirical testing against the new MSSQL engine.

Re:SQL Server = Almost Free

[Acer500]

I'm trying out MySQL's own administrator now, although admiteddly I haven't done anything besides adding a few permissions I needed:

http://www.mysql.com/products/tools/administrator/

I still find Microsofts' Enterprise Manager to be excellent, very intuitive.

I had to use Oracle's Java equivalent at uni a few times, and it was painfully slow and ugly (I've been told it has been re-done).

Re:SQL Server = Almost Free

[Cyberax]

MSSQL also has a LOT of problems. For example, its pessimistic blocking system very often gets in the way.

PLSQL is also somewhat nicer than TSQL.

Re:SQL Server = Almost Free

[kpharmer]

> Um what do you mean about the lack of command-line features? SQL Server has only one interface and that interface is SQL text sent to it from a client.

Sorry, I should have been more clear. I was referring to the limited batch processing ability of windows: that is, windows/2000 doesn't have a reasonable shell built-in. So, to automate tasks you're stuck with the windows task scheduler, primitive bat scripts, etc.

On most sql server projects I ended up installing cygwin, and running batch processing via shell scripts and python. Eventually we changed some very large applications over from using DTS to using BCP and python - with a huge and immediate reduction in maintenance effort & failures.

But most sql server dbas that encountered these systems were shocked - this just wasn't "the way" to run sql server, and these dbas just wanted to use the gui - and were obstacles to really getting tasks automated. However, the seasoned oracle, db2, etc dbas immediately recognized what we were doing and had zero problems in transitioning over. And of course when we later converted from sql server to postgresql most of the platform-independent code survived just fine.

Re:SQL Server = Almost Free

[Zebra_X]

"An experienced DBA can set up MySQL with many useful tools in a matter of minutes, too. And you can pay him more because you're not paying Microsoft."

In the grand scheme of things - the amount of money that goes to MSFT is insignificant compared to the revenue of a business. 5K for SQL Server standard edition - and "free" developer tools to support the DB. Almost every other database platforms packaged tools (i.e. provided with the default install) are insufficent to effectively manage AND develop for a novice or intermediate developer/dba.

Compared to other platforms the default toolset included with SQL Server (Enterprise Manager and SQL Query Analyzer now SQL Server Management Studio soon to be moved to VS.NET) is the most comprensive (though DB2 might be more so, but I have not used).

Finally, you don't NEED a DBA to setup SQL Server. Any reasonably proficient developer can figure it out, you also don't need a dba to tune sql server or to setup backup schedules or to create maintence plans. The traditionally cryptic dba functions are so accessible via the UI that there is really nothing to "learn" apart from knowing that you need to do X, and finding it in the menus.

As an aside - these TCO arguments also do not discuss which platforms are most productive for developers, this is where the most savings are to be had.

Re:SQL Server = Almost Free

[Pfhreakaz0id]

Umm, you do know that sql server has it's own scheduler for running scripts and stuff? SQLAgent? Obv. I guess not.

Re:SQL Server = Almost Free

[Serveert]

"TNS Listener"

Oh god, those two words give me the shivers, now I will have nightmares.

I love my pg_hba.conf.

Re:SQL Server = Almost Free

[plopez]

Postgresq installs are easier, IMO, than SQL Server installs.

I have done mySQL, hard to get 'right' as you need set all the options right for locking and data types for example.

SQL Server vanilla flavored installs are easy, but properly securing it, i.e., having it run as a user with lower privileges than 'System' is *hard*. You usually end up having to give it SysAdmin like powers which is wrong wrong wrong. Then there is the upgrade and patch tread mill and for any DB of any moderate size (greater than a few gigs) you need to buy a big honkin box to get enough memory cache as the OS and the DB engine can fight over memory. And low end boxes often don't give you enough slots for memory.

Oracle default install is easy and secure. Setting up table spaces is not more complicated than SQL Server IMO, though the cock pit layout is a bit different. It can run in a more secure mode than SQL Server. But you also end up on an upgrade treadmill.

Postgresql default install is easy. Tuning it is about the same as Oracle, though the cock pit layout is different, and setting up database files is easy enough. It installs to run in a sandbox which can throw you for a loop if you are expecting full access to the db engine right after install and can't connect to it. I.e. if you are used to SQL Server and mySQL.

So basically, those are my opinions. The main problems people have is that they are not flexible enough to learn a new cockpit layout. All of them can work well 'out of the box'. It is when you need flexibility that scaling where Oracle and Postgresql really start to shine.

Re:SQL Server = Almost Free

[LordEd]

Any reasonably proficient developer can figure it out, you also don't need a dba to tune sql server or to setup backup schedules or to create maintence plans. The traditionally cryptic dba functions are so accessible via the UI that there is really nothing to "learn" apart from knowing that you need to do X, and finding it in the menus.

As with any Microsoft product, it is POSSIBLE to do a lot of things without full knowledge of all of the functions. However, it is always better to have knowledge of what and how those functions are doing so you can take the best features from the product and use it to its full capability.

That said, I like SQL server. At work, we are starting basic experimentation with the mirroring system in 2005 to hopefully add some fail-over support to our network.

Re:SQL Server = Almost Free

[emdeejay76]

The points you raise are only valid for entry level database requirements. As somebody who has spent many years with both SQL Server and Oracle, I can tell you that as your demands increase the tool support provided by SQL Server, and indeed the database engine itself become critical scalability issues.

For example, the provided tools such as Enterprise Manager and DTS make routine data management simple - provided you have modest enough requirements that having a human berform the work is acceptable. Additionally, the data import/export facilities are terribly inefficient unless you run them on the database server (involving either remote desktop or scheduling the package). For large quantities of data import/export you simply must use the bcp command line tool, which has an interface so arcane it makes the Oracle command line tools look advanced.

Once you've either developed or purchased enough add-on scripting functionality to automate such tasks, you'll find that once your data volume reaches a certain point that the pessimistic fine-grained locking model, plus the naive query optimiser will have incredible impact on both your query performance and your concurrency. You'll find that many data management jobs will require 'staging' intermediate data transformation in temporary tables to work around the inefficiencies of the engine. Then, you'll realise that time is money and invest in a much more capable product.

Don't get me wrong, SQL Server is a nice product as long as your requirements are modest enough, and in a primarily Windows based shop it's often the most sensible choice. However the points you raise about Oracle (or any other enterprise level solution for that matter) are only relevant because your using a sledgehammer to squash a fly. When you're dealing with large scale data management situations, SQL Server is a toy.

Re:SQL Server = Almost Free

[jacksonj04]

However, even with PostgreSQL, setting up these things is more than an apt-get.

Re:SQL Server = Almost Free

[cruachan]

I agree with all you say, I used to work as a Oracle dba managing an enterprise level database so yep, MSSQL doesn't compare at that level.

However I've worked as a consultant/developer for the past few years, and I love MSSQL. As a Departmental or SME database it's great as it offers all the power and features I'd expect from a real database (as opposed to MySQL ;-) but takes virtually zilch maintenance when installed and configured. So long as I've got the backups and maintanance tasks in place I can deploy and a regular monthly check is more than sufficient. It may be a toy at the Enterprise level, but below that I really can't fault it.

Re:SQL Server = Almost Free

[Anonymous Coward]

There's plenty of command line facilities in SQL Server. For instance, type OSQL at the command line. You can create very complex SQL Scripts to perform all maintenance operations, such as backup, shrinking, truncating log files etc. You can use SQL Server agent to schedule tasks and creating recurring tasks which will inform operators on failure by a variety of means so I can't understand what tasks can't be accomplished at the command line.

You can create DTS packages and run them via the command line OR get the SQLServer Agent to run them.

As for copying DTS packages from server to server - http://www.codeproject.com/useritems/MoveDTStoAnot herMSSQL.asp

Are you sure your DBA doesn't spend all day reading slashdot?

Re:SQL Server = Almost Free

[WarwickRyan]

I've had experiece with MySQL, Informix and MS SQL Server as both DBA and developer. Recently I've dabbled a little with the free version of Ingres.

Of those, SQL Server's the easiest to use. Followed by MySQL (with phpMyAdmin), Informix (with gui) and Ingres. SQL Server and MySQL are simple enough that even someone who's not too familiar with databases (they'd still need to understand the relational model, mind) can use them.

If you're using your database as a simple datastore, with all of the logic being in a connected application then go with MySQL. It's a days work to install and learn all most people'll need to know. Which is massively cheaper than licencing of the other databases.

Overall, SQL server the most cost effective option when you require functionality over normal SQL, but under that provided by say building a java application. All thanks to the functionality provided by Transact-SQL. It's not perfect though, I'm of the opinion that Microsoft should move closer to VB/VBA here (C# would be nice, but too difficult for some MS DBAs).

An example would be DB warehousing: you can easily script everything you'd need to do to clean/import data from disparate sources in TSQL, without the expense of building an application and abstraction layer etc etc.

Oh, and SQL Server also has fantastic documentation out of the box, with two excellent books available that tell you everything you'll ever need to know (and make you an excellent DBA): Inside SQL 2000 and SQL Server 2000 Programming.

Re:SQL Server = Almost Free

[hey!]

Really? My sense from working with SQL Server is that the tools are mediocre. They're easy to use, if you limit yourself to absolutely routine work. Try doing a forensic analysis of the transaction log without third party tools. Oracle's tools are typically more complex than needed, but at least they're there.

I've studied the cost issue too, and in an apples to apples comparison, Oracle and SQL Server are essentially the same. You can spend a lot more in Oracle by choosing the wrong product (e.g. Enterprise instead of Standard) or the wrong license model. But Oracle has been very careful not to be undercut on price. Ethically, the Oracle Corporation leaves quite a bit to be desired, in that they tolerate selling very badly chosen license packages to customers, and leave the customer no option but to pay their way out. But Microsoft's lock-in strategy isn't any better.

Any experienced DBA can set up a database in minutes in any modern database system. And you don't need the command line , although I can't see that an experienced DBA should be scared by the command line. Personally I prefer using command line utilities when working with production databases because I can write down what I want to do, put it in a file, test the file, then execute the file knowing exactly what is going to happen. If you do things in a GUI, you have no guarantee you do them the same way twice. If you are an app developer, you have no guarantee that the customers will do things the same way twice either, which means a lot of long and costly support calls.

Speaking of an app developer's standpoint, MS-SQL is dreadful for developing off the shelf apps, unless you (a) use Microsoft IDEs exclusively and (b) limit yourself to the kinds of thigns the IDEs automate for you. The documentation looks extensive, but it is in fact very poor. The way to do many things are undocumented, except through the enterprise manager (see note about command line above) and then poorly. Some syntatical element aren't properly documented. Features are often poorly documented, and many "examples" are so trivial as to be useless.

The Transact SQL dialect is the least orthagonal SQL dialect in widespread use. You have to refer to the manual all the time in order to figure out what works where. The implementation of SQL is primitive; it wasn't until recently that the parser properly understood how to handle aliases; it's still extremely buggy when you use intermediate to advanced features like nested queries.

SQL Server is a good choice for outfits who (a) have a 100% microsoft policy, (b) aren't doing anything particularly challenging from a scalabilty standpoint, (c) do not require more than basic database capabilities, and (d) are only supporting internal users.

Re:SQL Server = Almost Free

[Anonymous Coward]

Right - nothing like running scheduled activities on multiple different schedulers:
    - Task scheduler for backups of file systems,file system size checks, periodic directory cleanups
    - sql server scheduler for database activities

Scheduling activities within a variety of products, rather than a single centralized schedule sucks. It especially sucks when they're extremely primtive and can't schedule down to the minute or handle the kinds of simple details than cron handles.

And it also sucks to use non-database functionality within a database. You've got schedulers, that horrific exchange email interface, crappy DTS ETL solution, etc - it's all garbage. They lock you into a single product worse than any mild sql dialect differences.

Re:SQL Server = Almost Free

[LWATCDR]

As I said it is a little bit not a lot. I tend to install the latest version so I have tend to have to use a tarball.

Re:SQL Server = Almost Free

[Hornsby]

I agree with most of what you said; however, any self-respecting Postgresql or Mysql installation definitely needs to be tuned to run decently. Postgresql in particular has the most conservative default configurations of any SQL server I know of. That being said, once it's tuned, the sky is the limit. SQL Server auto-tunes, which can save some time with initial setup, but it's caused me far more grief than Postgresql in daily use, and yes, I use both daily.

Re:SQL Server = Almost Free

[djbckr]

The lack of command-line features meant that many operational activities that could be automated required a dba to manually do the job via the gui

You're joking, right? The GUI executes the command-line interface. You need to learn how to use SQL Server

Re:SQL Server = Almost Free

[jedidiah]

He could also simply be confusing the database and the bundled tools.

          A number of the bundled tools in Oracle need to be "worked"
around. I could see someone used to DTS having coniption
fits over the default Oracle tools. Although I would expect
a group of competent developers to quickly adapt and route
around those limitations in an easily supportable manner.

          If you can't pick up a new language (bash,perl,C++) as
needed and abstract the sticky parts of a new platform (oracle)
then you aren't much of a developer.

Re:SQL Server = Almost Free

[jedidiah]

a) The TNS listener requires ZERO configuration.

b) Configuration of the clients is just a matter of dropping a file in place.
      If you are too dumb to do the pattern matching required to figure out what
      3 things will ever change in a typical TNS entry then you can just use the
      wizard and clone it's output.

Talk about helpless. Is this a generational thing or something?

Re:SQL Server = Almost Free

[Anonymous Coward]

What's all this talk about learning the commands?? Most database CLIs use SQL. If you're a DBA and you don't know SQL, then just go jump off a bridge. You can't handle learning a simple structured language? Find a new profession.

Re:SQL Server = Almost Free

[Anonymous Coward]

"In small businesses when you have one person who is your network admin / systems admin / dba / tech support it is important to have products that are easy for a (skilled) tech to figure out."

And then again, that' quite a lot about perception.

1) Microsoft products and, in general, closed software products are far far away from being didactical. That means is more dificult to master them (if at all).
2) Most of your "microsoft skilled techs" doesn't know their face from their ass. They only seem skilled since you are even more ignorant (not to seem rude here: you better should be skilled within your bussiness goals, why the heck should you be skilled on IT, if you are not IT staff yourself?).
3) Your only real claim is that -probably, it's easier on so many areas to find more people *claiming* their are Microsoft gurus (they aren't: I still have to find *one* IT professional that is skilled that their abilities came only from Microsoft products).

Returning to point 2, since you are not IT-skilled you are open to be fail on your percieved ability from an IT guy; I'll give you a (free) advise: specially since you need a "multi-hat IT guy" go for someone with a Linux background; as a rule-of-thumb he will be top notch compared to your average Microsoft-only guy... probably *even* on Microsoft-related tasks.

"but what's the learning curve on MySQL's CLI"

Does it matter? we agreed you are looking from an outhouse "expert", so the learing curve is at his expend. And that learning curve comes with a gift: you become more generally-knowledgeable from the experience (remember my advise?)

"every time I want to create a database I'm going to be wanting to right click on something and select the 'create database' or 'attach database' option... not try to remember if the command..."

That it's true, Microsoft knows it and that's why Microsoft products are the way they are. Of course, the lower front costs come at big expenditures on the long path, or do you think Mr. Gates became the richest man in the world by chance?

"So how does the TCO line up when you compare what your existing tech can do with Microsoft SQL, vs. paying $250 an hour for the _ONLY_ contractors in town who understand *nix or trying to bribe a DBA from the city to move to the country and work for you?"

The TCO lines (usually, there're always exceptions) brilliantly. With your "savvy" Microsoft "experts" you will just try to survive for as long as your bussiness is running; you apparently more expensive *nix guys can make marvels out of your bussiness only for a slightly higher front costs.

There were a very interesting (and quite old now) saying from Greenspun:
-On Apple it seems it will work, but somehow it doesn't.
-On Microsoft it doesn't seem to work and, of course, it won't work
-On *nix it seems it won't work, but the right *nix guru will make wonders.

Re:SQL Server = Almost Free

[Anonymous Coward]

"Any reasonably proficient developer can figure it out, you also don't need a dba to tune sql server or to setup backup schedules or to create maintence plans."

And the fact of your opinion being so common might have some relationship with the high ratio IT fiascos have reached specially in the last decade.

Re:SQL Server = Almost Free

[turbidostato]

Hey! your comment exactly mirror my experience on the *whole* Microsoft product portfolio from the last 15 years! (not that I'm surprised; if Microsoft policy has been one for about two decades and it made its owner the richest man in the world, why should they change it?)

Re:SQL Server = Almost Free

[Itninja]

Dangit! I must have proofed that post 10 times. And I STILL didn't get it just right....

So in other words...

[CDarklock]

Half of your database expenses aren't your software.

How is this a surprise?

[jimicus]

Unlike a desktop PC, any serious database installation demands a serious database and at least some professional expertise, even if it's just "sysadmin of many hats, one of which happens to be dabbling in the database".

Therefore personnel costs probably don't vary that greatly. This only leaves two costs: the application and the database itself. Generally speaking, the business will choose the application first and the database second (or they certainly should do), which means the cost (if any) of the application falls under the heading of "we've got to have it so it really doesn't much matter how much it costs, within reason".

This leaves the backend database, assuming there's a choice in the matter (not all applications support all databases, despite SQL being nominally independent). In such a project, licensing that is about the only really variable item in the list.

Re:How is this a surprise?

[RAMMS EIN]

``Unlike a desktop PC, any serious database installation demands a serious database and at least some professional expertise, even if it's just "sysadmin of many hats, one of which happens to be dabbling in the database".

Therefore personnel costs probably don't vary that greatly.''

Except that there may be a lower entry barrier to becoming an expert at one of the Free databases than to becoming one at a commercial database. I've worked with MySQL 3 and various versions of PostgreSQL, and, while I'm by no means an expert, I can at least develop applications that work and continue to work...on these databases. While I try to stick to features that are in SQL standards, and thus should be supported by all implementations (* casts an angry look at MySQL *), I won't bet any money on any of my applications working on Oracle. Even worse would be me having to maintain someone else's app, developed using the Oracle tools...I'd be a complete newb.

Re:How is this a surprise?

[laffer1]

There is a lower entry barrier just in that MySQL and PostgreSQL run on many more platforms than Oracle, SQL Server or DB2. Lets face it you can run those on Windows, Linux or Solaris depending on the product and then only on a subset of CPUs. Oracle sort of has a Mac OS X version but it sucks.

So if you like BSD or run linux on something besides x86 you are screwed. If you have a new shiny intel mac and only have a PPC version you are screwed. Some people can argue MySQL doesn't run as good as it should on non linux platforms, but at least it does run!

And for those linux users thinking so what if it runs on non x86, consider embedded devices. Imagine running a webapp off a PS3 or xbox360 or your cell phone. There are valid reasons to want products that are ported to new systems.

And don't forget the hidden touches

[monkeyboythom]

like must-have certifications for running, developing, and managing DBs.

Re:This just in!

[rossifer]

This just in, 0 is less than 100!

Seriously did someone really need to do a whole study to determine this and then write an article telling the world?

If you need clustering, failover, remote backup, separation of indices from data, you make heavy use of materialized views, and you need to retrain your DBA's on a new database, there are a lot of people who still say that it will cost you more to use Postgres than Oracle. Don't even think about MySQL.

But if you're starting from scratch on a new project and your current projections don't indicate you'll need a lot of those features, now the PHB's will have finally heard that free databases should be considered. We deployed on SQL Server and Oracle after developing on Postgres (because Postgres was about twice as fast when running the test suite). Postgres scaled better than Oracle on any single box configuration, and customer data sets never required more than 100GB databases in the worst case.

We were forced to deploy on Oracle and SQL Server because none of our customers thought that Postgres was enterprise qualified. Now, some of them might.

Regards,
Ross

Depends how you define business (lottery is not?)

[foobsr]

"German Lotto Company Plays it Safe with MySQL Cluster"

http://www.mysql.com/news-and-events/news/article_ 1188.html

And their application is not critical either, just win or lose.

CC.

Re:This just in!

[Anonymous Coward]

This just in, 0 is less than 100!

This just in, 50% of 100 is 50, not 0.

Whoops - reformatted...sorry

[scuba_steve_1]

Ditto this comment. When I rip out a simple system, I definitely lean toward Postgres by default...but when I need a system with any of the following qualities:

- clustering / high-availability failover
- robust WAN replication
- maximized performance
- ability to hire experts that can tweak every ounce of performance out of the servers
- extensive and complete documentation


...as well as range of other esoteric functions required for enterprise applications, I default to Oracle. Of course, this decision is also almost never an issue since each one of my clients is almost always already in possession of an enterprise-wide license agreement for the full Oracle product suite...because of Oracle's bulldog sales force. The downside to those licenses are, however, that I then have to explain why I would rather use JBoss than OAS.

Also, do not downplay the sheer number of people that you can hire that are Oracle DBA and PL/SQL experts. Licenses are cheap compared to the cost of a failed development program...at least in my world.

60% of..

[homer_s]

So what is the cost of commercial databases? Do the math:

Cost of OSS DB=$0 , which is 50% cheaper than commercial DBs.

0.5 * X = $0

X=$0

So, commercial databases really cost $0. I'm calling Oracle to get my copy.

(Yeah, yeah, TCO is not $0...)

Cheap Does Not Equal Reliable

[Anonymous Coward]

A cheap database is not necessarily a reliable database. Since an open-source (OS) database is not owned by any company, there is no final person who takes responsibility for any error in the database. By contrast, DB2 from IBM is a commercial database, and IBM guarantees error-free operation.

An OS database is appropriate for a non-commerical or minimally commercial environment: e.g., Slashdot. A commercial database like DB2 is appropriate for a heavily commercial environment like a bank.

Which database would you want to power your checking account? 'Nuff said.

Re:Cheap Does Not Equal Reliable

[Anonymous Coward]

So, because there is a company who is responsible if DB2 does fall over, that should make me happy?

If you have decent database administrators, there is no reason why you can't use OS databases. The fact that some third party can be held accountable for a failure doesn't make me want to use a bank using DB2 over a bank using Postgres.

And believe me, DB2 is capable of falling over. There aint no such thing as bug-free software. Nuff said.

Re:Cheap Does Not Equal Reliable

[pallmall1]

Which database would you want to power your checking account? 'Nuff said.
Gotta agree with you.

The simple rule of thumb is, "if it's got to be used by customers/clients, get it from a vendor."

Re:Cheap Does Not Equal Reliable

[CJSpil]

It certainly makes me happy!

No matter how skilled your DBA is there will always be something they can't fix! With an OS database that leaves you with the mailing lists (I'm aware that you can buy MySQL support) and if you can't track someone down who can help you with your problem then you're pretty much screwed.

DB2 and Oracle are perfectly capable of falling over, however the reason that the license and support contract are worth the money is the fact that if the sh*t really hits the fan then as a last resort you can always get one of their experts on site.

Re:Cheap Does Not Equal Reliable

[ultranova]

A cheap database is not necessarily a reliable database. Since an open-source (OS) database is not owned by any company, there is no final person who takes responsibility for any error in the database. By contrast, DB2 from IBM is a commercial database, and IBM guarantees error-free operation.

The world of databases must be very different from the world of every other kind of application, then, since every license I've ever seen has specifically denied any responsibility. Given this, I'd like to know if DB2 license guarantee means that, in the case of error, you'll get the license fee back, IBM pays some limited damages, or IBM pays all damages.

I find the last option quite unlikely.

Feature comparison

[phorm]

I haven't played with much DB work in the last few years, so I'm a bit out-of-touch feature-wise. My understanding is that in terms of not losing data, PostgreSQL was better than mysql, Mysql was faster, but as the move on they're evening out on those fields.

Is there a side-by-side comparison of major databases (including the open-source postgre, mysql) where one can see where in the featuresets each lacks/wins?

your mileage may vary

[kpharmer]

This is a pretty trivial article which seems driven by ingres.

Anyhow, a few things that I'd consider:

1. since as the author mentions the open source databases aren't ideal for mission critical applications (yet), then many organizations will find themselves supporting multiple databases. Say, oracle for financials & crm & the corporate warehouse and postgresql for a variety of smaller projects. Makes sense in many ways - except: oracle is already free for the small databases anyway, and now you need the dbas to support multiple products. This is going to increase your labor costs - not decrease it.

2. for many large analytical databases (data warehouses, etc) the cost of using open source are actually higher than closed source. This is because db2, oracle, etc are better at using the hardware than the open source alternatives. They've got better optimizers, parallelism, far better partitioning, better better pool management, automatic query rewrite, etc. So, a $100k oracle lisense running on a $100-200k 4-way (i know, assumes discount) will out-perform postgresql (free) on a 16-way ($1m) in many ways.

3. for some applications mysql could be more expensive than oracle. Ok, not just because you need to do far more testing with mysql to make sure that none of the wacky silent errors are affecting your code. But also because of the odd licensing - that requires its own faq and tips to just license the product if you can't figure it out. Then there's enterprise db - not very familiar with this one, but I doubt that it is free. Meanwhile, at the low-end the big-three database vendors all support free products. So, whether or not you pay more may very well depend on how you use the software.

Of course, if you're at a company like mine, and get to bypass purchasing and just review the license & install - you probably are saving a vast amount of money after all.

Re:your mileage may vary

[Anonymous Coward]

So, a $100k oracle lisense running on a $100-200k 4-way (i know, assumes discount) will out-perform postgresql (free) on a 16-way ($1m) in many ways.

200k for a 4 way ?
1m for a 16 way ?

A 4 way or a 16 way what ?

Check out X4600's from Sun - 16 Opteron cores and 32G RAM for less than 55k USD

Re:your mileage may vary

[glwtta]

1. Can be true - but depends on how small the "small" projects are, and how many there are. 15 small Oracle databases are going to be more work than 15 Postgres databases, even if you are already running Oracle for other stuff. Also, if there's enough work for the smaller stuff that it needs separate DBAs (even if it's just 2 total), then getting Postgres ones would not be more expensive than Oracle ones.

2. Blatantly untrue. Oracle will scale better on multiple machines, but on a single machine Postgres will outperform Oracle far more often than not. And I'd like to see some source showing Oracle's optimizer to be better than Postgres'. (and hey, depending on who's administering your stuff, that 4-way can outperform the 16-way with both running Oracle).

3. MySQL is shit.

Re:your mileage may vary

[newt0311]

just out of curiosity, has anyone ever published benchmarks for oracle, postgres, db2, mssql, and mysql in different system configurations (i.e. single core vs. 2-way vs. 4 way vs. 16 way) and in different load configurations (mostly serial read vs. parallel reads vs. read and writes with serialization)? I have never seen any benchmarks on this stuff yet and they would be very interesting to see.

Re:your mileage may vary

[kpharmer]

> 1. Can be true - but depends on how small the "small" projects are, and how many there are. 15 small Oracle databases are going to be more work than 15 Postgres
> databases, even if you are already running Oracle for other stuff. Also, if there's enough work for the smaller stuff that it needs separate DBAs (even if
> it's just 2 total), then getting Postgres ones would not be more expensive than Oracle ones.

I would only go with postgresql is it really was a separate team supporting them. Otherwise the oracle dbas are now going to have to also learn how to recover postgresql databases from backup, etc, etc. That just isn't worth it most of the time.

> 2. Blatantly untrue. Oracle will scale better on multiple machines, but on a single machine Postgres will outperform Oracle far more often than not.

Not for reporting. If I've got a four-way oracle server doing reporting with 100 million rows in its fact table partitioned on day with 2-3 million rows per day. Then oracle will be far faster at most trending (reporting) queries. You won't have 365+ tables to manage for fake paritioning, you'll get 4x the performance of the single threaded open source servers, and better memory management. It'll be *far* faster than the open source options.

> And I'd like to see some source showing Oracle's optimizer to be better than Postgres'.
Well, I was really thinking about mysql here - which has historically had difficulty joining more than three tables without unpredictable performance results.

Re:your mileage may vary

[glwtta]

Well, I was really thinking about mysql here - which has historically had difficulty joining more than three tables without unpredictable performance results.

Unpredictability seems to be ingrained deeply into the MySQL psyche. This always makes me chuckle: http://www.ex-parrot.com/~pete/mysql.html

Anyway, I don't think anyone is actually seriously pushing MySQL for the whole "Enterprise with a capital 'E'" thing...

I agree with the other stuff, Postgres does have extremely rudimentary partitioning abilities, but then it's not a difficult problem and will probably improve relatively quickly (ie, not on the same order as clustering).

Re:your mileage may vary

[quantum bit]

If you're paying a million for a 16-way you're an idiot. Partition your data and cluster cheaper machines. That's what Google does.

Re:your mileage may vary

[abirdman]

All of the commercial database software packages (that I know of) have a specific part of the license which forbids publishing benchmarks. You'll never see head-to-head benchmarks of MS SQL, Oracle, DB2, or any others, because they don't feel that any benchmarks would reflect the results of their product fully optimized.

In a sense, they're right. The "speed" of a database depends on many variables-- disk speed, network speed, CPU speed, contention from other processes, OS overhead, and countless configuration options-- and most vendors aim for (and achieve, in my opinion) "good enough" speed after applications and workloads stabilize and small performance tweaks are applied.

Right or wrong, the big software vendors don't believe a test scenario devised by a testing team, no matter how experienced or neutral, is going to reflect a real-world setup. What kind of workload is planned? Many small updates? A data warehouse/data mart where unchanging data is read often, but updates only once a day/week/month? Are reports quick reads of mostly unchanging data? Is this an OLTP system with many simultaneous changes from many users? Do queries have to be guaranteed to only return committed data? How much data will be involved, and what are the backup strategies? What are the connection interfaces? TCP/IP? ODBC? Java Beans? C++?

There's just no way to standardize the configuration of one single database product (that's why Oracle is legendary for all the configuration options), to say nothing of setting up several databases for a head-to-head test that's fair or particularly informative. Sad but true, it is unlikely that you'll ever see head-to-head comparisons of commercial databases, and the ones I've read of open source offerings are rarely that useful, because of the variability of real world requirements.

Re:your mileage may vary

[mitchskin]

All of the commercial database software packages (that I know of) have a specific part of the license which forbids publishing benchmarks. You'll never see head-to-head benchmarks of MS SQL, Oracle, DB2, or any others, because they don't feel that any benchmarks would reflect the results of their product fully optimized.

I have seen this said a number of times, and I believe it, but there must be some exception that allows the TPC results to get published. I'd love to see one of the Postgres support companies join TPC and get some numbers up there. They even have cost-normalized benchmarks (like transactions/second per dollar or data warehouse-type queries/second per dollar) that I'm sure would show the open source databases in a pretty good light. Especially in the "non-clustered" category. Even if Postgres didn't win, just having it shown in the same context would be great.

Re:your mileage may vary

[Anonymous Coward]

I couldn't figure out in your gibberish how Oracle is cheaper than MySQL, nor could I understand how you got wacky errors out of MySQL. I've never seen any. SQL works as it should. It links up very well with other software, including ODBC connections, etc. Again, I couldn't figure out from your gibberish what you are really trying to say. MySQL supports their product. You can support it yourself, or you can pay them for support. Its not really a very hard concept. If the software didn't work, people (eg NASA) wouldn't use it. ...btw, NASA wasn't hypothetical, they really do use it. http://www.mysql.com/news-and-events/news/article_ 51.html

Re:your mileage may vary

[cruachan]

I'm afraid you're the one with the gibberish. It's a common misconception that all databases are equal, it's just that some are larger than others. That's simply not true, when you get to multi-million row tables with scores or hundreds of tables in mission-critical environments then this is a quite different beastie to your department level database or even web site backend. Also, in one of these large databases the tables and data are probably considerably less than half the business functionality wrapped up in it, the various relationships defined, procedures, functions and triggers will contain vast amounts of business logic are just as important as the actual data. Furthermore having these coded up at server, rather than application level (aka MySQL) will save money in development and management that could offset TOC of even Oracle to below that of MySQL.

It's true of course that MySQL has been improving of late, to the point where they finally seem to have recognised that Atomicity is a good thing, relational integrity might have some uses and subselects can be fun :-). Haven't they even got some sort of transaction sql scripting now too?. Unfortunatly I still get the impression that these might be tacked on and they certainly don't inspire the sort of solid confidence yet that Oracle, DB2 or even MSSQL supplies.

Re:your mileage may vary

[turbidostato]

"There's just no way to standardize the configuration of one single database"

Yes there is, an easy one. Vendors know it, still they don't want to produce it.

In case the vendors were not aware, I'll explain here:

*Go to three universities with reknowned IT areas expertise. Ask them to produce ten "Enterprise database case scenarios" so you have thirty of them.
*Pass an Slashdot-like poll to Fortune 500 DBa's so they can not only vote but pose their opinions too. Go again to the universities with the poll results so they can refine their test cases till the point ten of them sum up 80% of the votes.
*Invite whatever DB "vendors" (including of course open source ones too) to a closed door test with limited budget (say one million dollars per vendor; valuate the DBAs man/hour costs at the average from the Fortune 500 ones an the hardware and software by its front cover retail prices). Then allow them (within budget) to produce results on whatever the vendor-provided DBAs considered proper (say, two Oracle gurus working on their preferred 256 ways Suns, two Postgres hippies and a FreeBSD greybearded on a cluster of white brand boxes, half a dozen Windows architects on Ms Windows Enterprise Datacenter on a bunch of 16 ways DELLs... whatever) give them a two week time frame and compare the results.

You would say you are comparing apples to oranges, but all the parties worked within the same budget and time constrains and where allowed to bring whatever experts they considered proper to achieve better results so from the Fortune 500 standpoint is the clearest apples to apples comparation you can get.

Still, they don't want you to know.

Re:your mileage may vary

[Anonymous Coward]

LOL, dude you are seriously screwed up

> couldn't figure out in your gibberish how Oracle is cheaper than MySQL,
lets see - the parent said that in a data warehousing or reporting scenario:
    - oracle software & 4-way server hardware cost could equal $300k
    - mysql software & 16-way server cost could equal $1m
what exactly is not to understand? Disagree that parallelism, optimization & partitioning are effective in this scenario? If you do, you really need to crack a book on data warehousing or reporting!

> nor could I understand how you got wacky errors out of MySQL. I've never seen any.
ah, then you've never really tested your mysql code. If you had you would have discovered that mysql has gone its own way with exception handling - and generally doesn't provide an exception if you give it data that doesn't match type requirements, is too long for field, etc, etc. It's what the entire "mysql gotchas" list is all about. Just google it for gods sake.

> If the software didn't work, people (eg NASA) wouldn't use it

Right, so based on this logic IE is a better browser than Firefox since more people use it. ActiveX is a nifty idea because NASA uses it. COBOL is a great language because NASA uses it. Big deal, just because Joe Schmoe at Nasa decides to use Microsoft Bob for his next project doesn't mean squat.

Meanwhile, do people a favor and perform a little research on mysql before you jump to its defense. And the whole "gibberish" accusation needs to go - at least until you get your hands around English grammer.

product support is over-rated

[kpharmer]

> People are far more dependable when they're working for money than for charity.

not when they suck - which they frequently do when working on product support teams.

yes, I'm glad that I'm working with supported products - but I also avoid calling them like the plague. It is very much a worst-case scenario.

Re:product support is over-rated

[MidnightBrewer]

I was thinking of the developers, not tech support. Of course, you have a point - just because somebody has credentials doesn't mean they didn't come through by the skin of their teeth on the job interview.

me = totally sorry

[marcello_dl]

here's the POT version :D

> Most open source database products, including MySQL, seem to require quite a bit of digging and cobbling together to set up and maintain.

Not on linux IMO. Of course if the windows port of postgres used to require, IIRC, cygwin it's not really a db fault is it? Never had problems with a couple of local and remote installation of Postgres. And the fcgi app connecting to it survives live updates of the OS and the db code.

>Microsoft SQL Server has fantastic tool support, no command line experimentation required.

Well I won't ever doubt it. My support tool is a mailing list which is not fine for enterprise.

A command line tool is very welcome, for example when depl oying to a shared server which needs the ram for the daemons, not for pretty GUI apps. Also, autocompletion and help of psql seems adequate to me. I often prefer it to firing up the GUI tool.

> An experienced DBA can set up a new installation in a couple of minutes.

Under linux the package system usually sets up a new installation. One just has to put something in the ACL of the db to allow the app an appropriately secure connection if needed. An experienced DBA sets it up in seconds, me in minutes :)

fluff piece.

[CDPatten]

Anyone every take a good long look at the price difference between ibm, oracle, and MS? MS really isn't in their price league.

"UP TO 50%". Ya, thats great, can you be a little more vague. Given that MS, Oracle, and IBM all have different prices, WHAT ONE WAS THE 50%? What conditions created that? We could include Centura SQL Base in that group and still keep the phrase up to 50%, but centura hardly breaks the 4 figure mark.

IBM and Oracle are way more expensive than SQL Server. You want to impress me, break it all down and show us REAL NUMBERS for each database software, then tell us who quoted them prices to you. Give us a little facts to back up your claims. Explain to us WHAT support costs you are factoring in, because I'm tired of seeing journalists who are damm near technologically illiterate write tech reviews.... On top of that they aren't even neutral... they are really writing op-ed pieces (like this one) and pass it off as a study.

You know, people on here bash MS & Oracle (or whatever closed source company it is) when they see a "fluff piece" about their products... I just hope they bash this fluff piece too, I won't hold my breath though.

Re:This just in!

[Angvaw]

I switched from Oracle to PostgreSQL. I miss summary management (materialized views, dimensions, query rewrite), various ETL features, streams, data guard, updatable views, multi-table inserts, log miner, flashback, database links, packages, automatic workload repository, OEM, and the several months of my life I spent working around PostgreSQL's lack of anything comparable to these features. (I'm not just listing random features - I've wished I've had all of the above at some point...) PostgreSQL is nice for what it is though - a solid, scalable open source database. And psql is much nicer than sqlplus (tab completion - woo!). And I don't think I've ever gotten the Postgres equivalent of an ORA-600...

Well its obvious...

[kiwioddBall]

The reason why enterprises don't like open source is because they HAVE been insecure and less feature laden. Anyone can write a database. It takes skill and $$$ however to write a secure database with enterprise features such as failover, 100% availabilty, hot backups, massive scalabilty over the planet, full support, and even more have had all these features PROVEN.

Being nearly there doesn't cut it at all. Being proven does. I wouldn't put my multi billion $ business on the line with some piece of free software developed by who knows with what agenda and debatable level of quality ... I'll pay for something, and have support and a legal avenue if it falls over.

Re:Well its obvious...

[Anonymous Coward]

Yea, that same attitude worked well ... just look at all of Microsoft's security!

Re:Well its obvious...

[Software]

I'll pay for something, and have support and a legal avenue if it falls over.

What, precisely, are your legal options if Oracle falls over? It's pretty standard in the software world (COTS or otherwise) to disclaim liability for every type of damage known to man, and to disclaim warrants of fitness for a particular purpose.

Re:Well its obvious...

[quantum bit]

Hahahahahahaha. Oracle has to be one of the least secure products (and companies for that better) I've ever seen. Far worse than anything MS ever put out. Just do a search on Bugtraq for people complaining about security bugs they've filed years (!) ago finally being patched, only to discover that the "fix" doesn't really fix the underlying issue but just papers over one particular vector of exploiting it.

Despite whatever marketing rhetoric Larry may spout, Oracle's attitude toward security is still what MS's was 7 years ago (la-la-la-I'm-not-listening!).

Re:Well its obvious...

[kiwioddBall]

Yeah, I've posted in the Oracle security FUD front page news on this. Its not a major issue with Oracle. You should read the post (check my other posts) but basically, Oracle back end servers are generally hidden behind so many firewalls that it doesn't matter, grater insecurity comes about from having client applications leaving usernames and passwords in freetext somewhere on the system than by buffer overflows etc. and there are greater security issues in all databases such as SQL Injection, rather than exploits.

Security is more than just patches and exploits etc.

Your argument is crap though - saying that a product is crap because it has issues doesn't make your product any better. A very common fallacy on Slashdot.

Re:Well its obvious...

[kiwioddBall]

There are real security issues with client facing products such as Microsoft Windows and IE etc. Back end products that are generally hidden behind lots of firewalls etc have a different range of issues with different priorities.

Again... just because you think MS's product is poor doesn't make Open Source any better.

Re:Well its obvious...

[kiwioddBall]

Yup, good point. Usually if you are using, for example, Oracle then you have some leverage with the company though since they don't want to lose the business, and it isn't unusual in the business world to add liability clauses as a condition to winning that business. Its not all in the license.

Re:Well its obvious...

[Anonymous Coward]

> Your argument is crap though - saying that a product is crap because it has issues doesn't make your product any better. A very common fallacy on Slashdot.

You'd like to think that was the point he was trying to make. What product, exactly, was he trying to "make better?"

Re:Well its obvious...

[bill_mcgonigle]

The reason why enterprises don't like open source is because they HAVE been insecure and less feature laden.

In what universe has PostgreSQL been less secure than Oracle or MSSQL?

No one was ever fired for...

[lukateake]

... selecting IBM.

It's a tired cliche to be sure but it's still true nonetheless.

I think it's fair to say

[Toreo asesino]

...for mission-critical operations, most people don't give a flying fuck about TCO, purchase cost, or any other cost - it's about proven & reliable track-records combined with bullet-proof support. Oh, that any high-performance scalability, fail-over clustering, and all the other things OSS databases haven't quite caught up on yet.

Remember kids, if it's expensive - it must be good right?!

Re:I think it's fair to say

[Cederic]

Except that if you chat to MySQL (the company) they'll tell you that they do have high performance scalability, fail-over clustering and all the other things. However, they're also keen to charge you a licence fee for that stuff, and support costs at the level you'd expect for qualified professional support staff.

So TCO pretty rapidly does become an issue.

Oracle lost a sale to us because despite having a product that would give us far better performance the TCO argument didn't work out. Frankly we'd rather just chuck another CPU and a few gig of RAM at the problem than give the money to Oracle.

I like Oracle, I want to use Oracle, but if they're 40% more expensive than the alternative (taking into account licencing, support costs, hardware, training, recruitment, maintenance overhead, etc) then I'm not doing my job properly if I recommend we pay that premium ahead of an alternative open source database that still meets our technical requirements.

Of course, the same applies the other way. I wouldn't recommend we use Postgres because we have no skillsets in-house, and recruitment would hurt, and it just doesn't give us anything our existing DBMS lack.

If TCO wasn't a concern we'd all get a pay rise...

"Not losing data" has to be mentioned?

[PRMan]

I don't think I could consider a database where "not losing data" is one of it's weaker points. I haven't lost data on a SQL Server since version 6.5. NEVER.

If I wanted a database that periodically lost stuff, I would just use Access... ;)

Re:"Not losing data" has to be mentioned?

[phorm]

I think you misinterpreted. The reference was not to spontaneously losing data, but rather due to a particular circumstance or failure. For example, many databases feature the ability to confirm, roll-back, and otherwise validity various transactions to a strong degree. Not if, for some reason, your server had an error/failure or a network glitch occured, you might lose data or get corrupt data. A good database may have ways to detect and/or correct this, whereas a lesser one might not.

Re:"Not losing data" has to be mentioned?

[quantum bit]

I don't think I could consider a database where "not losing data" is one of it's weaker points. I haven't lost data on a SQL Server since version 6.5. NEVER.

Don't tell that to the Russians...

In all seriousness, I think what the OP was referring to is MySQL's tendency to silently change input data rather than aborting the transaction with an error. Examples include things like invalid dates, or out-of-range numbers being truncated. It's been a major MySQL gripe for a long time. The latest version (finally) has a "strict" mode that enables checks for things like that, but I don't believe it's on by default. Not 100% sure what the status of it is as I'm a PostgreSQL junkie.

phpmyadmin is great

[SethJohnson]

Hey, I'd like to highlight the parent ninja's recommendation for phpmyadmin.

It's a very convenient tool for performing a lot of tedious command-line operations. Perhaps the one thing I've run into as a failure is when you try to export large data sets, the PHP renderer usually will timeout. That's when you gotta bust the command-line out and earn your keep. Phpmyadmin is also great for performance-tuning your DB. It shows you where the bottlenecks are and advises on what config parameters should be modified.

Seth

Re:phpmyadmin is great

[Bloater]

I'd recommend pgadmin3.

Re:phpmyadmin is great

[bigman2003]

I'd recommend that the guy who said he wanted simple installation and management does not want to learn the difference between the 3 MySQL management systems that were just mentioned.

Sometimes having choices just complicates things. He can install SQL Server and just tell the other person, "Here, use Enterprise Manager" without having to think about it.

Re:phpmyadmin is great

[jedidiah]

I don't think that guy actually wants to "manage" anything. He just wants to run the installer and forget about it.

Re:phpmyadmin is great

[Bloater]

That's perfect! postgresql comes with an installer in the download. Sounds like postgresql would be ideal :)

why innovate when you can replicate?

[Anonymous Coward]

The initial outlay for a licence is only relevant to the smallest businesses...there is no other reason whatsoever to choose mysql/postgres over, say, Oracle. If you want a database application that is actually well tied to the architecture (especially in clustering terms) of the underlying platform, RDB on VMS is God; unfortunately, HP's business dream is to become another Dell, but with a wider range of printer cartridges.

Read the article

[Mycroft_514]

It says that the 50% only applies to databases of sizes 200GB and less. Well guess what? Not only is my companies systems 24x365, but we have 1+ trillion rows of data, spread out over 1000 tables. 200GB? I have that in less than 10 tables!

And that is just the mainframe. Now add up all the UDB and SQLserver databases, and we probably have another 500 million rows spread out over those systems.

Re:Read the article

[Almahtar]

This doesn't apply to me, therefore is irrelevant. Can't you all see!??! What's wrong with you people?!?

Re:This just in!

[newt0311]

"I miss summary management (materialized views, dimensions, query rewrite), various ETL features, streams, data guard, updatable views, multi-table inserts, log miner, flashback, database links, packages, automatic workload repository, OEM, and the several months of my life I spent working around PostgreSQL's lack of anything comparable to these features."

PostgreSQL has query rewrite and updatable view thanks to its rules system. the others are missing (to my knowledge).

SQL Servers can be pretty expensive

[@madeus]

Oracle typically costs hundreds of thousands of dollars to run on an enterprise DB server like an E20K or E25K. The commercial version retails for around, if I'm not mistaken, tens of thousands of dollars per CPU (and these things can have up to 75 processors in them, though anywhere between say 8-24 is closer to what most have in them) and most companies with an Oracle DB on them are going have at least two systems.

Microsoft SQL Server is indeed cheaper than Oracle for commercial use, but it requires you run Windows (which you also have to pay for - not forgetting relevant per-client license fees), which is a dubious enough prospect on it's own if you have a commercial service relying on it - the real kicker is Windows only retails for x86 systems (i.e. only suitable for low to mid range use). Even the X4600 systems (which support up to 8 dual core AMD64 CPU's, and indeed are insanely fast) are not going to be suitable big iron DB server replacements any time soon.

An experienced DBA can set up a new installation in a couple of minutes.

Well given the cost of the system compared to what's freely available anyway, I should hope it would be quick.

Installing it and getting a DB up and running still takes longer than this does though:

#apt-get install mysql-server
#mysqladmin create my_database
#mysql
mysql> GRANT ALL on my_database.* TO user_account@localhost IDENTIFIED BY 'p4ssw0rd';
mysql> exit;


You can install MySQL on a Mac with only basic motor skills in your right hand in about two minutes (just double click the icon, click "Next"->(etc)->"Next"->"Done"), and it has a GUI control panel to start and stop it too, so you literally don't need to open a terminal window.

The real cost is in the time spent compensating for whatever your database platform's tool support shortcomings are. I love Microsoft SQL Server for this reason: I rarely have to reinvent anything.

I completely agree with the sentiment that having tools is a worth consideration. Apart from the pretty good CLI interface on MySQL (one of the reasons I think it has wider adoption than PostgreSQL, which is less user friendly), I use CocoaMySQL, though phpMyAdmin is more popular of course. There are also Windows utilities, and a range commercial tools from MySQL AB. There are similar offerings for other databases like PostgreSQL.

Personally I only tend to prefer something liek Oracle over a free, open source database when I know I need good performance and a reasonable level of reliability and I have to store a metric truckload of data (I work in the telcoms/large network provider industry so it's not uncommon), and coming up with a way of doing the same thing with something like MySQL or PostgreSQL just isn't worth the overhead.

I actually try to avoid using something like Oracle if I can, not because I have a problem with the product (and not because of cost) but more because the more Oracle databases you have, the greater chance some muppet (9 times out of 10, a consultant, who knows bupkis) will break it performing routine maintenance.

I think something that's worth bearing in mind is you don't always need to use a database, even when using one seems like the obvious thing to do. This might sound odd to some people, but it's becoming increasingly common to consider other approaches, now people are being asked to build systems that need to store ever more data. Basically, you can store you data in much the way as large service providers have been storing mail and web content for years.

For example, say you have to store a record of a list of transactions going through a system. Could be authentication logs, could be B2B data, something generic (it's not an approach suitable for everything, but it is a very flexible method of storage).

You might be better off using a disk base format like this:
/data-storage-

What database from Microsoft?

[cyman777]

What database from Microsoft do you mean? Database? Microsoft? No, that does not fit together... :)

Re:What database from Microsoft?

[Anonymous Coward]

don't worry this is slashdot and people will excuse your complete ignorance of the Database Market as long as you have a dig at MS.

FYI, SQL Server is the fastest growing market share DB in the industry, has a security record infinitely better than Oracle and has many of TPC-C and TPC-H price performance records and unlike the OSS equivalents it supports clustering and failover extremely well.

Re:What database from Microsoft?

[imemyself]

If you're going to bitch about M$, atleast bitch about something that's true. MSSQL Svr is actually a pretty damn good database. It may not run on anything but Windows, but as long as it does what it needs to do, it really doesn't matter.

Access sucks balls, but then nobody with a clue uses Access anyway.

And don't get me wrong - there are *plenty* of good things to bitch about as far as Microsoft goes. Like the disorganization and inconsistency of Office 2007 RTM...

Re:What database from Microsoft?

[cruachan]

As others have said, SQL Server is a stunningly good product, and if Windows and Office were written to the same standard then nobody would be bitchin' about Microsoft.

As someone who has used SQLServer, Oracle, DB2, PostrgresSQL, MySQL and several other smaller DBs on various projects over the past 20 years I've been in IT, and has been DBA on some of those too, SQLServer is most definatly my favourite. It's absolutly solid, runs itself largely and has excellent features. I also like Postgres, which can stand with the heavyweights. MySQL is most definatly the worst: although it has been improving recently to the stage it can be considered a database rather than a flat file management system.

Re:What database from Microsoft?

[Anonymous Coward]

that's rough. your first post in months, anti-MS just to be safe, and it gets unanimously trashed by SLASHDOTTERS?

Smells like utopia

[Caesar Tjalbo]

I'd so love to choose the db. But noooo.
"Here's the COBOL manual." "Whut?" "We don't have a license for Fortran anymore. Oh, and we're behind on the documentation because we were going to migrate to ."
So OK, that was the worst example. Where ever I go, I always encounter legacy databases that have to be worked with. That and an Access 'thing' lovingly setup and maintained by dr. Clueless which managed to wurm itself into the production process.

Where are the test results?

[msobkow]

Do you seriously think any CIO with a functioning brain cell is going to go with free unsupported software when they can't even find a single reference to such databases from any certified performance evaluation companies or organizations?

The downtime cost of one single failure in a five year period for a mission critical system can easily run 100 times the cost of a commercial product with support. Only bean counting fools risk their entire business without properly assessed risks and disaster recovery plans.

Not having someone to source the recovery of the smouldering crater that was your data center is a huge issue.

Re:Where are the test results?

[neelm]

Because when it goes down, it costs money. When it goes down, I throw more money at it. It goes down often, so I throw more money at it to keep it up. Sometimes, just because I think it might help, I throw money at it. And when they tell me were loosing money, I reply it because we haven't thrown enough money at it.

I thought databases ran on computer systems, but I'm learning they really run on money.

Re:Where are the test results?

[msobkow]

Databases run on computer systems.

Businesses run on legal and financial systems.

A viable business deals with it's business needs first. Technology is an expense to meet those needs, including risk mitigation and protection from lawsuits over failures to meet contractual obligations.

Get your head out of the sand. Business runs IT; IT does not run business.

Re:Where are the test results?

[neelm]

It's a good thing then that IT departments run so well, or I would think there was a fundamental flaw in that argument. We've seen time and time again how open source product are no threat and can just be ignored. Best left to those little companies worried over irrelevant details, like consumer satisfaction. We'll focus on the important things, like law suits and blame.

Re:Where are the test results?

[msobkow]

North America, Europe, the UK, and other regions actually demand that you take fiscal responsibility for delivering services. If you can't accept that, you don't run a business here, you're just waiting to be stepped on by someone's lawyers.

Open source is a perfectly viable option for applications with mitigated risk. It is also technically viable for business. But it's your insurance company and legal department that determine if it's a viable business move -- and they like numbers. Hard numbers. Not geek speak banner waving pride.

Re:Where are the test results?

[neelm]

Not arguing the point - it is okay if people die as long as you had your legal disclaimers in that people may die within the acceptable margins. I get that, it's totally cool.

Re:Where are the test results?

[msobkow]

*snerk*

Good flamebait. Irrelevant, but amusing.

Re:Where are the test results?

[timjdot]

Not sure of your implication but I think you are saying Open Source is not mature. First, MySQL was not written by 19 year old kids. Secondly, Posgresql is in lineage older than Oracle of Microsoft's RDBMS. Thridly, most of the Internet runs on Apache and probably most of it runs on MySQL much more than any other RDBMS. Google on google adwords and MySQL. They did not have to revert to MysQL from Oracle because of problem in MySQL.

What I have always seen is some techie or psuedo-techie manager specs out Oracle or a Miscrosoft product. Then the company buys it. The techie is a specialist in something and has been to weeks of indoctrination training in Oracle or Softie so obviously will recommend whatever he/she thinks will give the importance, high pay, and longevity. The manager read a magazine once and went to a tradeshow and wants to pick the system that doesn't rock the boat.

Fortunately, the current generation of techies have used open source. But it takes alonog time to get rid of dead wood. I consulted in 2001 for a huge insurance company which still required 3270 screen scraping and input rather than opening up their mainframe to JDBC. Well, that inefficient process had worked in the 1960's when the now managers/execs started so they knew it and were comfortable with it. Most of all, they had insurance contracts for at least 1 entire state, maybe 1/4 of the nations military and poor, and such. They were set. Nobody is going to compete with them because the choices are political rather than practical. This is the present issue with Open Source. While mega-corps fall bechind theis opens a real opportunity for the SMB's to use the best technology and leap forward.

If you are still recommending Oracle or Microsoft's SQL Server or even db2 then you owe it to your client to get up to speed on MySQL, Postgresql, and other Open Source projects. Of course, if yor client is profitable because of politics or marketing and not because of core value, then you can recommend smoke signals if you want.

BTW, do you even have any clue what the Oracle contract requires WRT publishing benchmarks? Suffice to say, you can't unless they first approve. At least that was the deal back when they did they $1M BS challenge in 1999/2000. You're mention of a CIO with one brain cell is right. Any CIO still buying Oracle must be smoking dope. Plain and simple. (At least unless they have a large installed base of apps using non-standard SQL etc). That reminds me of a story one of the other consultants told about how they walked in on their CEO and he was smoking dope in his office... but that company only had a few $10-$20M J2EE projects it was doing for clients!

Have you ever studied set theory? Have you ever studied computer science? An RDBMS is a real software applciation. No amount of BS or FUD can remove the real engineering and science behind it. Fortunately, this science is well-understood and also mature.

In responce to your implication that an Open Source package is not supported and that it would take down an enterprise, I call "BS". Have you even placed a call to MySQL AB to talk to them about this? Of course Oracle is larger but that's because they are selling a 50 cent belt for $50. The era of globalization leverages mass manpower and Open Source does as well. Why fund $80M submarines and such rather than the ones actually doing the work? You'd do much better for yourself and the world to spend money with an Open Source company than with a proprietary one. At least from the practical viewpoint of future product development.

TimJowers
http://www.serviza.com/ : Serviza Monster Linux Computer with Open Source Training Bundles

Re:Depends how you define business (lottery is not

[Anonymous Coward]

I lived in Germany for two years. Trust me, their not that smart.

To be fair...

[HangingChad]

...it was based on interviews. You'd have to read the original report for the methodology. TCO savings can be calculated by project or enterprise function. You can't dismiss the conclusion as fluff without knowing the background of the study.

I can say on a project level OSS database products are almost untouchable on a cost basis. At least every one I've been involved in.

Re:To be fair...

[cruachan]

Yeah, but that sort of misses the point in you can't really compare them on straight TCO. If I was looking a multi-million row enterprise mission critical database for a large organisation then really I'd stick with DB2 or Oracle. Coming down the scale to department of mid-sized company level MSSQL or Postgres would be top of my shopping list. Smaller again and we're into SQLServer Express, Postgres, or in some circumstances MySQL.

Of course the lines are blurred and products tend to improve over time, but generally I think you'd find most developers with experience of a wide range of databases and companies would tend to agree with the above in outline.

Laughable!

[gbutler69]

Where are these fictional DBA's who know how to write advanced SQL or procedural code? I've never met them. All the DBA's I've met are marginally competent clowns. Of course, so are most of the software developers I've met (aside from Open Source).

Re:Laughable!

[jedidiah]

"advanced" huh?

That's just a telltale sign that you're fixated on using features just because they're there. In all likelihood you will make it more likely that your code is unreadable, unmaintainable and does unpredictable things while tending to exercise arcane features of the product (any product) and triggering obscure bugs.

Re:Depends how you define business (lottery is not

[Anonymous Coward]

I lived in Germany for two years. Trust me, their not that smart.

Something must've rubbed off, because the correct spelling would be "they're" as a contraction for "they are".

Re:This just in!

[amanda-backup]

>>> If you need .... remote backup... Don't even think about MySQL.

Remote backup of MySQL is available:

http://zmanda.com/backup-mysql.html

IOW- Free isn't free, you save 40-60% and.....

[CFD339]

...for that 40-60% you give up something.

How much you give up depends on your needs of course. It still leaves plenty of competitive space for commercial products however. Is support, higher end functionality, and formalized business practice worth paying 40-60% more for? In many cases it will be. It remains up to the vendor to make the case. Is their higher end functionality really "higher end"? Is their support good enough to justify it's cost? Is that corporation surrounding your purchase adding value?

Open Source is just another kind of competitor. Commercial software vendors want to keep Open Source software as a low-end option only, and avoid it being considered competitive at the high end. Since they can't change what the OSS community does, their only choice is to continue to innovate at a faster pace that can only be supported by all that money. Sounds good to me.

Ah, capitalism remains safe from the invading threat of commie open source once again. My faith is restored.

Re:IOW- Free isn't free, you save 40-60% and.....

[antirelic]

When IT budgets reach a certain point, TCO is less of a concern and liability/blameability is more important. Open Source databases can handle pretty much anything, and there is no lack of expertise in these database systems on the market (for hire). Most large companies want a phone number to call when the crap hits the fan, and they also want to know that these companies have been around for a long time, and will be around for a long time. Most CIO's dont care that everything you need to know is on google, unless you can call google and a google technician will fix the problem over the phone, or in person due to a lavish platinum support contract. Can CIO's be the big problem in the world?

Summary

[jbellis]

kiwi: "The reason why enterprises don't like open source is because they HAVE been insecure"

quantum: "Actually, Oracle has a demonstrably worse security record"

kiwi: "Security doesn't matter. Also, I'm rubber; you're glue."

Lesson: STFU when the facts pwn you.

Re:Summary

[kiwioddBall]

What is it about Slashdot that attracts people who can't read and don't listen, trolls who decide they're right by the amount of abuse they can deal out?

Fact : Enterprises don't like open source because it has been insecure, isn't proven, is developed by possibly dubious people and doesn't have the features they need. End Fact.

Deal with it. Abusing people doesn't change the fact!! :)

Re:Summary

[dcam]

I have a proverb for you:
It is better to keep your mouth closed and let people think you are a fool than to open it and remove all doubt.

Free versions from the "big guys"

[davidbrit2]

If the cost difference we're talking about here is simply the licensing/upgrading cost, it's worth noting that several of the popular "mega expensive" database platforms offer free (as in $0) versions - albeit with certain functionality removed.

Microsoft SQL Server 2005 Express Edition
Only supports databases up to 4GB, and is lacking the built-in task scheduler, and most of the high-availability and business intelligence features, but is perfectly usable for small-to-mid-sized applications/web sites. Plus you can upgrade later to one of the fancier versions if necessary.

Oracle 10g Express Edition
I haven't had a chance to play with this yet, but it looks similar to SQL Server Express in terms of features and limitations.

IBM DB2 Express-C
I don't really know anything about this one. I just now found it in a fit of "I wonder..." The product comparison pages don't really say much about it, but they'll send it to you free on a DVD, so that's pretty neat.

Sybase ASE Express
Never used this one either. It seems to be only for Linux.

Though honestly, from what I've seen of Postgre, I'd almost think that one would be worth looking into more so than these for small systems. One of these days I'll get around to experimenting with it. The advantage with the Express Editions is, however, that you don't have such a nasty learning curve if you can just jump right in with a database platform you're familiar with from at work. Why else would I do something insane like running php + MS SQL Server? :)

Re:Free versions from the "big guys"

[Anonymous Coward]

Sybase ASE Express
Never used this one either. It seems to be only for Linux.


The express (free) edition is only for Linux but there was talk of them releasing a free edition for all the Windows and Unix platforms they currently support. Also considering ASE was in fact the original "SQL Server", moving from MSSQL to ASE is much easier than any other cross db move. So TCO and ROI are somewhat complex things - if I were to move from MSSQL to Linux/ASE, I end up saving on OS and platform and dont need to compromise on enterprisey features.

When I can more easily move to a free and stable database for small footprint uses, why would I want to risk things with a move to an OSS db when all these other databases are tried tested and proven in very demanding environments (like stock exchanges and the government). This isnt a flame, I'm seriously curious!

Re:Free versions from the "big guys"

[Anonymous Coward]

The free version of Oracle is easy to install and works great, and comes with lots of additional stuff to learn on. Highly recommended.

Re:This just in!

[Angvaw]

I was referring to materialized view query rewrite, where if a MV can answer your query and the MV is fresh enough for your stale tolerance level, the MV will automatically be used instead...

Good point about the rules system though.

Forbidding benchmarks

[etymxris]

I would never rely on any software that forbid benchmarking.

Re:Forbidding benchmarks

[abirdman]

You can benchmark to your heart's content. It's publishing the results that get you into trouble.

PostgreSQL support rocks too

[einhverfr]

Pervasive exited the market recently saying that the email lists were just too good to compete against.

The issue is what sort of talent pool you can attract, not what sort of companies back it. And PostgreSQL has an awesome talent pool of developers.

Note you still don't have (and will probably never have) an equivalent to RAC, but that is something maybe a proprietary company can come in and offer? I have ideas on how this could be done but I don't think it would be doable as an open source project...

Re:PostgreSQL support rocks too

[Anonymous Coward]

A number of companies like CommandPrompt, EnterpriseDB, and even Sun Microsystems offer support contract for Postgres. The latter two even offer 24x7 enterprise class support. All three companies seem to be working hard with the community to advance and grow the postgres community. It rocks! The MySQL stuff seems to be a lot more closed. Yeah there are a lot more ISVs but I like the mailing lists, community and the environment around postgres. Also the comment about really low admin is spot on with the right contrib stuff it is way easier than both Oracle and MySQL.

Re:PostgreSQL support rocks too

[Anonymous Coward]

"I have ideas on how this could be done but I don't think it would be doable as an open source project..."

Why the heck not?

Re:PostgreSQL support rocks too

[bill_mcgonigle]

Note you still don't have (and will probably never have) an equivalent to RAC

Why would open source never have this?

but that is something maybe a proprietary company can come in and offer?

Is this like RAC?

Re:PostgreSQL support rocks too

[einhverfr]

The problem is that it would be applicable to a small niche market and would actually bias the software against the lower end by reducing performance and increasing complexity.

Re:PostgreSQL support rocks too

[einhverfr]

Note you still don't have (and will probably never have) an equivalent to RAC
 
Why would open source never have this?
 
but that is something maybe a proprietary company can come in and offer?
 
Is this like RAC?
 
  No, Greenplum's MPP offers some share-nothing clustering capabilities for sections of a dataset. It also offers parallelism within queries hittind different nodes on a cluster. But this is different.

RAC offers true shared-everything clusters. THis means that you have shared block devices, shared processing servers, and the like. It is fairly complex and involves moving coordination out of the system itself and into some sort of distributed lock manager (usually part of a clustered filesystem like OCFS2).

Now, here is the problem. PostgreSQL is a very strong application. I don't think it is trivial to compete against them as a fork in the open source world, and the PostgreSQL community is not interested in doing what would be necessary for shared-everything clusters. THese things would make PostgreSQL less competitive in the majority of the markets that the community is in at the moment by adding unnecessary I/O overhead, and added complexity of the code would slow down development. There are some other sticky issues too (disk-based DLM's are going to be slower than in-memory semaphores, for example).

So, what you are left with is something which won't have the broad interest that would make the product less generally applicable. So it isn't going to happen.

I would compare the situation less with Green Plum (where there would be interest in their control node code) and more with EnterpriseDB. EnterpriseDB's big thing is Oracle compatibility. PostgreSQL doesn't aim to be compatible with Oracle, and so a lot of stuff like handling empty strings as NULL's (or is it the other way around) in Oracle will never be accepted into PostgreSQL. Hence EnterpriseDB has to maintain a set of changes to the parser to do those things that the PostgreSQL development community for good reason doesn't want.

Re:PostgreSQL support rocks too

[einhverfr]

Agreed. I work with Josh Drake of Command Prompt, Inc. on a daily basis. His company has been doing awesome things with PostgreSQL support. Josh Berkus from Sun also is involved in some of my projects (see the LedgerSMB link below and look at our core team).

But the support on the lists is incredible. Of course, one would probably not rely solely on the lists for that mission critical application and would want someone who one can call to get things fixed immediately.

The thing is, Pervasive thought that if they just offered support, they could come in and help corporations adopt PostgreSQL. They were wrong because the early stages in the adoption were happening with support primarily from the mailing list (and a *lot* of support that may be very complex, but isn't time sensitive, happens on those lists). They threw a lot of time and money without really understanding to what extent they were competing with community (and free) resources. FOrtunately, CMD, EDB, and others are not making that mistake.

Re:PostgreSQL support rocks too

[bill_mcgonigle]

Thank you. Great explanation.THese things would make PostgreSQL less competitive in the majority of the markets that the community is in at the moment by adding unnecessary I/O overhead, and added complexity of the code would slow down development. There are some other sticky issues too (disk-based DLM's are going to be slower than in-memory semaphores, for example).Could you elaborate on how Oracle manages a RAC and non-RAC mode but PostgreSQL couldn't do so? So you'd write your locks to disk in "RAC" mode but keep them in memory is standard mode... no? I have to assume Oracle does the same or they'd have been suffering the same fate you outline above since their support of a RAC mode.

Re:PostgreSQL support rocks too

[einhverfr]

Could you elaborate on how Oracle manages a RAC and non-RAC mode but PostgreSQL couldn't do so?

I don't know the internals of Oracle, but I suspect that all interprocess communication probably goes through some lock manager abstraction. Note also that Microsoft SQL Server has no equivalent to RAC, nor does Firebird/Interbase. In fact the only other system that I can think of that supports shared-everything might be DB2.

I see challenges (which Oracle seems to struggle with as well) in:

1) scalability v. performance on the low-end (Oracle sides with scalability, as you will note if you try to run Oracle on a smallish server or a new-ish desktop with only 1GB RAM).
2) performance on high-end v. simplicity (Oracle again sides with the former).

Let me describe to you the issues involved here:

Most databases find that they can increase performance by keeping all coordination between database operations in memory, local, and as inexpensive as possible. This means that many systems use in-process memory and threading to accomplish this (MySQL, MS SQL), while others use semaphores and per process architectures (PostgreSQL). What happens here is that different processes have to serialize their access to the block devices so that they can make sure they don't trample on eachothers' writes.

Operating systems tend to cache some storage information in memory through the standard disk cache. Because the mappings of the files to the storage are among the most frequently used bits of information, they are also usually cached in most filesystems (including both NTFS and EXT3). THis means that the computer assumes that it knows exactly what is going on regarding the block device. If something from another operating system writes to the block device in the mean time, the operating system's information is out of date and bad things happen (like data loss, filesystem corruption, and the like).

Even with a clustered filesystem, you have an issue here: if you don't coordinate reads and writes properly, you can corrupt information within a specific block or page of the db table. If you assume that two computers could be accessing this at the same time, this becomes hard. In fact, you can't run your coordination through a single computer and have to rely on DLM's. This means a performance loss. Also since this is a very common function, abstracting this is not going to be inexpensive either. Perhaps this is why only Oracle and DB2 offer this option. MS SQL, Terradata, PostgreSQL, and others aren't generally interested in the small segment that needs uptime measured in decades. So in these cases, going with the DLM-type solution would actually be a bad thing.

Here is the thing: Shared-everything clusters offer unparalleled reliability but at a performance and complexity cost. Because they are applicable only to a small section of the market, I cannot see that one could generate enough interest in this sort of thing to sustain an open source project. These patches would never be accepted upstream either.

Hope this helps.

Re:Atheism responsible for 50% of all genocides!

[Anonymous Coward]

+1

you must be kidding

[oohshiny]

MySQL is absolutely trivial to set up; on most Linux distributions, setting it up is a single command to install the package, create the databases, install the service, and start it up.

There are standard GUIs for administering MySQL, probably for people like you, but most people I know prefer the command line. The command line is easier even for non-DBAs, because it's far easier to write down and document what needs to be done.

An experienced DBA can set up a new installation [of Microsoft SQL Server] in a couple of minutes

Probably true. But with MySQL, you don't even have to be a DBA at all, and it's probably even quicker.

OS = free = you get what you pay for

[sinij]

Lack of support, poor backward compatibility and shifting standards are all common problems for open source.

I worked for the company that used MySQL for very data-intense large application and my recollection was that a) updates always broke something b) some things never worked as intended and required hacks c) it cost money to get developers to fix specific bug that severely affected them but very few other people d) problems with fail over, where any hardware failure caused few hours delay from real time for a day or so. I'm sure overall cost, even with costs of time lost by database programmer, was a lot less than enterprise database.

what utter nonsense

[old_guys_can_code]

What company with a $150B market cap uses MySQL to store their mission critical data? One that doesn't take their advice from IBM...

postgres works great!

[turing_m]

This article should mention Postgres.

In upgrading a mission critical MS Access database, I first looked at MySQL, then after a month more of research I ended up going with Postgres (aka PostgreSQL). It will always be free due to the BSD license. It does what a database should do, supports things like referential integrity without the need for hacked on tables ala MySQL, and has few gotchas. Seems very quick to me.

And did I mention free? If you have a boss who is a real tightwad and doesn't understand computers let alone databases, Postgres is a godsend. Especially if he'd rather pay your salary than Larry Ellison. Some bosses would rather pay $100k in employee salary than spend $10k on software. I'm not saying that Postgres will take $100k worth of employee salary to figure out, I'm just making the point that there are bosses who only like to fork out money for tangible things, like hardware or employees. If you need to accomplish something at work and don't want to go through the stress of going hat in hand to your boss to buy something he will not understand and begrudge you for, Postgres is just what you need.

The market leaders have had to put out a free intro product to compete. So what. Eventually your needs may expand, and then Ellison or Gates (or MySQL) start holding you up by the ankles to see how much money drops out. Postgres will still likely be happily chugging along. Any features it lacks will probably come online by the time you need them anyway.

So, you might have to spend a month or three learning all the setup and maintenance aspects of Postgres (and Linux or FreeBSD, if you are unfamiliar with them). The good thing is, you only have to learn it once. Then you are up to speed and you should be as effective as with any other RDBMS. Now you can roll out highly scalable, rock solid internal business applications. I'm not even a CS major. I'm just a EE who has taken 3 or 4 CS classes, none of which involved databases.

Not everything is amazon.com. It may not be glamorous, but there are huge numbers of small - large businesses and departments struggling away using Excel for things that need a database, dealing with questionable data and lots of double entry (or more!) due to inherent problems with flat files. Sure, you could use Access, but it's a bit like buying a goldfish bowl for your new pet baby shark. Or you could go with MSDE or whatever "free" intro thingy Oracle has. But why would you bother? Postgres does all that and scales to most conceivable sizes for such businesses.

I've found the support to be very good, from both IRC and the mailing list. Just make sure you do some googling, RTFMing and possibly buying one of the existing books out there.

In fact, I think that Postgres really only lacks a couple things. One is a really good book. The other is the name, which is confusing. If you need to google for info, PostgreSQL gets the most hits.

Re:postgres works great!

[turbidostato]

"In upgrading a mission critical MS Access database..."

And then I stopped reading.

They becoming cheaper by the minute...

[jonney02]

From the title to the story clip they became 10% cheaper...:) *duck and run*

Re:No one was ever fired for selecting IBM

[SolitaryMan]

Still want to meet a person, who was fired for selecting Postgres/MySQL/Firebird

Actually, this is the good hint for boss on how a person does her job:

• IBM/MS/Oracle -- Plays it safe
• Postgres/MySQL/Small, unpopular product -- really knows what she's doing

Not necessarily true, of course.

re: clustering

[toadlife]

It's my understanding that clustering is much improved in MS SQL 2005 over 2000. Do have any experience with it? Is it still much inferior compared to Oracle's clustering ability?

I'm the "DBA" (I use that term loosely) where I work with SQL 2000, and am just curious.

Fluff statistics

[Archtech]

Whenever I see a phrase like "up to 50%", I always rephrase it mentally as "not more than half". Funny how much difference that can make to the overall impression you get.

Reality

[Stormcrow309]

I am about to tick a few of you off, but oh well. I work at a hospital as a system analyst. Look at my profile for my qualifications. We tend to purchase systems that are best fit for our environment, then write the glue to make them work. We have had systems running on MySQL that we have purchases the last couple of years. Every single one of them have converted to another DBMS package, usually MS SQL Server - with one moving to Oracle. Every system we have had running on MySQL has been undependable and unreliable. Support from the vendor has stunk and we end up having to troubleshoot the database issues without much help from the developers, because they do not have DBMS vendor support. When they switch to a commercial package, support quality improves drastically. All a person has to do to understand why is to go out to MSDN or Technet. Microsoft isn't the only vendor that has such fantastic support, but they are a good example.

The support issues with applications using MySQL have gotten so bad, that we now have a directive from my CIO that we are not allowed to implement any system based off of MySQL. The cost-benefit is not there for us to try to keep a MySQL database stable.

Re:Reality

[Anonymous Coward]

Ah, I bet you were running MySQL on windows... I have found that MySQL on win32 is not stable at all, however on Linux or Unix it is rock solid.
Companies need to get off of the Windows money sucking bandwagon and use a real server OS that can have uptimes of over 1 year.

Re:Reality

[Stormcrow309]

Unfortunately, you have no clue what you are talking about. Redhat Enterprise 3. In addition, one can't say we don't have sufficient in house support. We have significant DBA support: 4 qualified in Oracle, Sybase, and Microsoft. A server support group with Linux, AIX, HP-Unix, and Windows background. AIX and HP-Unix guys (who also do Linux) have 10+ years experience each in Unix. We have programmers who between them cover a good portion of modern programming languages in common use. One can't say we don't know how to deal with data load. We deal with systems with up to 7 GB of growth a week.

To handle a simple medical forms program (not truly that simple, since it adds the patient info to the form so it has to have an HL7 interface), with about 300ish form types and about 1500ish users, the MySQL database couldn't perform. The vendor couldn't get the app to scale properly across a two-system cluster due to lack of developer support. The MS rep heard from one of our DBAs and the vendor was offered a lucrative package to convert. We payed for Win 2003 and SQL Server 2k enterprise(with a sharp discount), with an uptime movement from 78% to 99% (we had a broadcast storm from a core switch). In clinical applications, even just creating and printing forms, we need the crap to work.

I don't care about if MS is better then OSS or Big Iron or Swiss Cheese. If the system can't get the job done, I have no use for it. If it takes the capabilities beyond what my 140+ IT department has to run, then it is useless. We run a ERP, Patient Accounting System, Electronic Patient Records, and a myriad of other systems on different platforms. The only OSS based system we have is our security system, which works great. Ideals don't matter when saving lives.

Sorry to all the normal people for the rant. It has been one of those days.

Google MySQL

[avapex]

Although there is limited support, MySQL can be clustered on a load balanced system. Just search the MySQL site.

Google embraces open source development with MySQL being their database of choice. I haven't found any articles about Google using any proprietary database management system.

Re:Google MySQL

[Anonymous Coward]

> Although there is limited support, MySQL can be clustered on a load balanced system. Just search the MySQL site.

mysql's clustering sucks - it's an in-memory database that they bought from another company. It's only mysql in name - a completely different codebase. So, if you've got a 100 mbyte database it may work for you. If you've got a twenty gbyte database it won't.

> Google embraces open source development with MySQL being their database of choice. I haven't found any articles about Google using any proprietary database management system.

They started using mysql back when there weren't many cheap/free options. Who knows what they use today for sure? And who cares? Unless of course, like them you've got an effectively unlimited budget for R&D. MySQL has a variety of problems that have always increased the cost of development for database applications beyond the trivial.

From the research that I've done it appears that google has learned to appreciate data modeling & architectures - I would be surprised if they're still building these databases using a product from the company that claimed 99% of applications don't need transactions or referential integrity.

There's no excuse for anyone at this time to use an inferior product like this:
    - the recent slashdot loss of threading for hours a couple of weeks ago - caused by mysql allowing the user to configure indexes differently than columns in a table. Then it required a twenty million row table to be re-indexed. Which took 4+ hours. On any other database that problem couldn't have happened, and reindexing a database with just 20m rows would have taken just 10-20 minutes.
    - date fields allow dates like Feb 43rd
    - database may silently truncate strings and numbers
    - you may specify innodb but it may built tables on myisam if it hits a snag - and it won't tell you
    - you set up strict constraints (the only option in all other databases) but any client can turn them off and insert crap into your database
    - no partitioning
    - no parallelism
    - no clustering
    - immature codebase
with compelx licensing in which you may or may not have to pay, and they may or may not change the terms next year.

So, if you've got options like Postgresql, or free versions of db2 and oracle - why bother with mysql?

Not a Vendor Supported Option

[dwandy]

Often times it's not the 'enterprise' that's demanding a commercial RDBMS: It's the off-the-shelf application that demands db2, oracle or ms-sql ...no mysql or postgres in the list of 'vendor approved' databases.

And even though you think it *should* work (sql is sql, right?...heheh), note that lots of these apps query the system catalog, which means they _need_ to know how the catalog can be queried, which means you need to pick a support DB.

If my-sql and postgres spend time getting included as an 'option' for commercial products then perhaps even if they don't get picked the first time, it would increase their mind-share and make them an option next time...

Re:Not a Vendor Supported Option

[somersault]

very true.. one of our accounting apps requires MS SQL server, and currently uses some 5 client lite version, but I'm thinking about getting full SQL as the timesheet system I recently wrote was done with MS SQL Developer Edition. If we get full SQL then I may as well do any corporate web apps with MS SQL. I was going to use mySQL, but after reading this page I think I'll go for postGRE..

Research by Forrester

[symbolset]

That's incredible!

Forrester? The intellectual giants who brought forth Rob Enderle and Maureen O'Gara (The MOGTroll)?

This has to be solid analysis then! They wouldn't just make up meaningless numbers to compare the incomparable. That would be disingenuous, disreputable and intellectually dishonest.

So who paid for this study?

But the open source databases generally lack the features for mission critical applications, trailing behind their proprietary peers in security, uptime, performance and features such as XML support.

Ah, that would be Oracle or Microsoft then.

whats so bad about mysql??

[Anonymous Coward]

http://www.mysql.com/support/

MySql5 is fast, stable, secure, reliable, supports clusters VERY well and you can buy support just like any of the "Big Boys". My company developed a claim management system for Insurance companies and its ALL on mysql. And yes, we have a full time DBA, any BIG app running off DBs needs one. For us what was the difference? Well, licencing fees during the first 2 years of development for one. Now whats the difference? There is none. We still dont pay any licencing fees. We still have a DBA. If we were runing Oracle or SQL$, we would still need one, maybe have to pay him more, and probably have to constantly send him to really expensive training also...(Any DBAs want to comment on that one?) Im not saying Oracle is bad. its an entity. its almost alive... if you're not running on the TOP-OF-THE-LINE server, its rather slow.... SQL$, thats another discussion....

Anyways enough of my blah blah blah.

The defacto is this; In the end, we must go with the Tried and True products for mission critical systems. Well, Google uses mysql, hows that for tried and true?

Why kids quit?

[HikingStick]

Most high schools simply want cookie cutter kids. Teach them all the same stuff (with little variation allowed), the same way, and push them out on the other end. There is rarely any accommodation for differences in learning style, or for a young person's unique talents, aptitudes, or approaches. That's what made me hate high school, and I hear the same from young people today, too.

I have received praise in every job I've held for my communications skills--verbal and written. When finally in college, my writings were nominated numerous times for recognition and publication. In high school, however, my grades were mediocre. I lost points for not having all of the requisite drafts for term papers, even though I could sit down without a formal outline and write a first draft that was better than the final papers of most of my peers. I failed junior year English composition, because I did not complete a paper on what I wanted to be when I grew up (I would still have a tough time with that one today, almost 20 years later!--I still don't know what I want to be). That missing English credit (and some other poor marks where I was penalized more for thw "how" of my work rather than the work itself) kept me from receiving a high shcool diploma, prevented me from going to college full time for many years, and directed me on a very unpleasant path for a number of years.

Math teachers automatically penalize students who solve problems in nontraditional ways (or without showing all of that long multiplication or division)--a real problem for intuitives who could derive the answer by looking at the problem. I'm one of those (but probably not as gifted as many other /.ers). In time, however, I learned the value of showing my work (to help others in the process, or to allow others to proof my work), but to force young people into tedium for tasks that are second nature to them makes them despise school. The only reason why I did not drop out is because I thought my parents would kill me (only a slight exaggeration here).

Later, when I returned to college as a nontraditional student, I was given much more latitude in how I approached my work, I was able to select courses that better fit my interests, and my grades were top-notch. Unless high schools are willing to break the cycle of disfunction that treats students like cogs on a wheel (after all, aren't modern high schools simply extensions of the industrial revolution?), more and more students will become disenfranchised and will risk dropping out.

which is what I referred to ....

[CFD339]

...as "formalized business practice". It absolutely has some value. As does support contract availability, and advanced technology.

Again, as I said the value and reality of each of those is measured by the buyer. For some, support contracts and formal businesses hold no value. For some, the so called "advanced technology" represents vendor lock in. It is all about perspective and need.

FSM bless the free market system.

Fear and Loathing in

[Anonymous Coward]

Fear sells,

The fact is fear will sail a product into the open streams of the world market. If its a sure thing.
Cost is considered appropriate to education. To put this in a simple formula, millions for the software : thousands for the individuals. Trust the individuals or trust the machine.

A 19 year old computer enthusiast is questionable in terms of his ability to take orders. A 40 year old developer is reliable. The 19 year old will take his pay and spend it on necessities he sees fit. The 40 year old will invest in non-essentials such as investments(c). Too many times has the old and wise sold us short for the pleasures of life.

Seriously? Who cares. If one is to start a company one has to be mindful of his product.

Know the tools first. (mysql source code)
Know how to repair the tools (mysql source code)
Know the environment (GNU, Linux)
Know your customer

How can you honorably represent yourself when you know nothing of your toolset other than how to use them. A warrior that didn't know something of making weapons, didn't live very long.

Re:Depends how you define business (lottery is not

[Anonymous Coward]

I love the guys who come to these forums to correct spelling and grammer. They all type with one hand...

Enterprise ready Open Source DBMS

[alderX]

Check out MAXDB [1] (no it's has nothing to do with the regular MySQL stuff - except the name - the engineers are payed by SAP and sit in Berlin). It is solid, scalable, has a decent speed. It has all the enterprise features like triggers etc. and is very easy to administrate and use. It has some nice features like no need for reorgs, no tablespaces (just data files/raw devices in which the tables are stored in) etc.

It is out there as open sources (GPL) since 2000.

[1] http://www.mysql.com/products/maxdb/

Another point on "50%"

[msobkow]

If half of your software purchase, deployment, and maintenance costs are consumed by the software license, then you pay your IT staff an insultingly low wage or simply don't bother with little things like disaster planning, growth projections, tuning, or anything else that makes a database efficient.

Any organization that is spending half their IT expense on licensing is a joke. The only way that claim makes any sense is if someone is rolling in service and support and claiming it's part of the capital expense instead of an ongoing run-rate service charge. Fire your accountant.

Wrong topic?

[RedTheRat]

This might belong to the High School article. Anyway, I found a few typos: thw, shcool - which is not consistent for a person so self-confident of its written communication skills.

Also, I think you are generalizing from your specific point of view a little too much. For me, the tight rules and directions at my high school fit my character much better and resulted in better grades than the laissez-faire at university.

Re: clustering

[Vancorps]

I'm not up on my licensing yet for SQL 2005 but replication and interoperability is much improved. Clustering SQL is now on par with Oracle 9i although 10g has much better management over the process while being harder to setup. That's always been the trade-off for me when it comes to Oracle or MS SQL. MS SQL is easy and fast to deploy, but Oracle scales better and performs a hell of a lot better with the same hardware. It's also harder to setup. Oracle took a backwards steps with 10g. It is much harder to setup than 9i was but it's still quite manageable with the documentation readily available.

I've found when it comes to Oracle, be prepared to read. When it comes to MS SQL, you can usually get by just looking around the UI until you find what you're looking for. It is a much more friendly environment. Firsthand, I have a SQL distributor talking to an Oracle database and 2005 makes it easy as pie. I like them both a lot although they both have their faults. Oracle seems to have come to their senses with licensing which is still the major roadblock for Microsoft.