Linux Helping Oracle

Mr. Fahrenheit writes "CNN has a story about how Oracle's effort to port their database to Linux may be helping them to out pace IBM." From the article: "In its biennial survey of the world's largest databases, WinterCorp, a database research and consulting company, reported that Oracle dominated its list of 175 large databases. For the first time, databases running on Linux appeared on WinterCorp's list -- and all of them came from Oracle."

Comments

Larry Ellison

[mnemonic_]

Truly an American icon.

Re:Larry Ellison

[Anonymous Coward]

as in greedy, self centered, and thinks money can buy good taste ?
yeah truly an American icon

Re:Larry Ellison

[Nutria]

as in greedy, self centered, and thinks money can buy good taste ?

As in had a plan, built a company, worked hard, saw it to fruition and made a lot of money.

Yes, an American icon.

The same America that foreigners have flocked to by the millions for 150 years, because their native countries are too fscked up.

Re:Larry Ellison

[IdleTime]

All I can say is that after living in USA for the past 10 years, is that USA is more fucked up than any other country I have visited or lived in and at the moment, the count is up to about 40 countries.

It is always Americans without any form of international experience who come with these allegations, and i begin to understand why and the reason why is not very pretty.

Re:Larry Ellison

[Nutria]

All I can say is that after living in USA for the past 10 years, is that USA is more fucked up than any other country I have visited or lived

If it's so fucked up, and you're not a citizen, you've been here for 10 years because?????

Re:Larry Ellison

[Silverstrike]

Probably because the economic incentives to live here are hard to ignore.

That says absolutely nothing about the sociological climate though.

I'm a citizen, and this is my home, so I would never leave, but a majority of the people in this country are seriously f**ked up.

Creationism anyone? That is considered a serious debate in "These United States". You do realize that the rest of the world looks at ideas like that and laughs at us, right? Not because we have people here at think like that -- all countries have their share of crazies, but because there's enough of them to actually cause a political movement -- in a LARGE democratic country.

Re:Larry Ellison

[IdleTime]

Easy to explain.

The company I work for brought me here because of my abilities and skills. Even though I have a green card, I will not be living here for that much longer. And if it is of any comfort to you, I can make 3 times what I make here in the US back in my home country, but I value the challenges of this job more than money since I have more than enough for a comfortable life. Add to that that the climate here in Florida is way better than back home. Looking at a home down in Keys at the moment so I can do some fishing too.

Luckily I have a great pension and a great health care system in my home country so i would never trade my citizenship with a US one.

Enlightened Self-Interest

[ThatGeek]

This story demonstrates enlightened self-interest, not kindness. While us folks using Linux will get a better operating system as a result, the Oracle corp will get an OS which can run its software well.

I just wish the people at ATI and NVIDIA would start to understand that giving technical details to open source developers doesn't always have to hurt. Technology is not a zero sum game; it's like science in which there are benefits to working together.

Re:Enlightened Self-Interest

[ScrewMaster]

And you know what? I'll take enlightened self-interest over kindness any day. At least, then I know where I stand. But if someone is being kind to you for no apparent reason, you really have to wonder about their actual agenda.

Re:Enlightened Self-Interest

[Anonymous Coward]

Makes you wonder what those guys with the "practice random kindness" stickers are REALLY up to.

Re:Enlightened Self-Interest

[aussersterne]

Spoken like a mean person.

Re:Enlightened Self-Interest

[Ruie]

And you know what? I'll take enlightened self-interest over kindness any day. At least, then I know where I stand. But if someone is being kind to you for no apparent reason, you really have to wonder about their actual agenda.

I see you have chosen your nick "ScrewMaster" for a reason..

Re:Enlightened Self-Interest

[ScrewMaster]

Read what I said again.

The term "enlightened self-interest" has nothing to do with screwing the other guy ... in fact, it very much has to do with deliberately not jacking the other guy around. That's what the term "enlightened" means in this context. I ran a consulting company for about fifteen years, and yes, I was out to make a profit. After all, that's why I was in business. But I fully expected the other guy to come away from the table with some benefit as well, which is what any good business relationship is all about. If your only goal is to get whatever you can, by any means, no matter what the cost to your business partners or your customers, well, that's self-interest without the enlightened part.

Furthermore, when people would come to me with "opportunities" that seemed to good to be true (i.e., being "kind" for no apparent reason) I was naturally very suspicious. I would always ask, gee that sounds great ... but what do you get out of it? If the answer was nothing I knew they were lying, and that I would end up getting the shaft. But if someone came to me and said, "I have a mutually profitable business arrangement I would like to discuss with you" I would at least listen, because they were being honest about their expectations.

Re:Enlightened Self-Interest

[danratherfoe]

This story demonstrates enlightened self-interest, not kindness.

In my view, the reason why ATI and NVIDIA do not release specs is not because of "lack of enlightenment", it probably has more to do with the fact that they are infringing on each other's patents. I'm not in the graphics hardware field, but it is my understanding that it is impossible to build a product without infringing multiple, multiple patents. When you release specs, your infringements are aired for the world to see.

Re:Enlightened Self-Interest

[danratherfan]

Did you consider the possibility that if they don't release their specs Bill Gates will give them head? Just a thought.

Re:Enlightened Self-Interest

[kfg]

As a practitioner of random acts of kindness I understand where you are coming from with the sentiment. On the other hand I know where ScrewMaster is coming from as well.

With someone looking out for their own enlightened selfinterest you know the rules. You can talk. You can negotiate win/win situations. If something goes bad and you get screwed you can still talk and negotiate to figure out why and what can be done about it.

There is no talking to someone being "kind" for "your own good." These people are likely to leave you broken and bleeding on the ground and walk away with a warm fuzzy feeling about it and a selfcongratulatory smile on their faces.

And promoting true kindness is an act of enlightened selfinterest.

KFG

Re:Enlightened Self-Interest

[Anonymous Coward]

More like: If they release their specs Bill Gates will have their heads...

Users of non-Microsoft platforms are also wary of Microsoft gaining additional leverage on the makers of graphics hardware, who routinely support both OpenGL and DirectX. In Microsoft's US antitrust case, the company was found guilty of using illegal monopolistic practices to discourage PC manufacturers from promoting software, such as Netscape's Web browser, that competed with Microsoft applications.

Re:Enlightened Self-Interest

[ScrewMaster]

Actually, they could easily (if they haven't already) cross-license their own patent portfolios to eliminate the problem of infringing each others patents. That's what all the big boys do, to avoid being drawn into expensive and fruitless patent suits. But, yeah ... the more sophisticated (and successful!) your product line the more likely you are to infringe, particularly in this day and age. I mean, suing over patent infringement has long since ceased to be a matter of protecting an inventor's limited monopoly as it is a profit-making business model in itself.

Re:Enlightened Self-Interest

[0racle]

How do you know he doesn't work for a hardware store? Meanie head.

Re:Enlightened Self-Interest

[lixee]

But if someone is being kind to you for no apparent reason

I believe you meant "some corporation" ...

Re:Enlightened Self-Interest

[ScrewMaster]

Yeah. Or some government busybody.

Re:Enlightened Self-Interest

[killjoe]

There is something profoundly wrong with the world and mankind when acts of kindness are looked upon with suspicion and kind people are berated.

Re:Enlightened Self-Interest

[ScrewMaster]

I wasn't picking on kind people ... I was picking on people who deliberately give the appearance of kindness, when in fact they are something else entirely. But yeah. There's something profoundly wrong with the world all right.

Re:Enlightened Self-Interest

[lawpoop]

You are right. With enlightened self-interest, both parties are capable of positioning themselves to benefit. With kindness, it can be taken away at a moments' notice.

Re:Enlightened Self-Interest

[Anonymous Coward]

You're blowing the whole thing way out of proportion. Either you employ coercion as the means to your end, or you employ voluntary association as the means to your end. The former is moral and fair; the latter is immoral and unfair. It's simple, unambiguous, and instinctively understood by every normal human being. Self-interest is irrelevant, since basically every decision you make in life is self-interested anyway. (Why on earth would you make a decision if you did not think it was the right thing for yourself to do?)

Another way of describing this is human nature. It doesn't take a rocket scientist to understand, because it was evolved that way.

Re:Enlightened Self-Interest

[ScrewMaster]

Not really. I think you misunderstood me. What I'm describing is sociopathic behavior, and the fact that sociopaths are a. often very, very hard to detect and b. are looking out only for themselves. People are surprisingly easy to fool, and the root of all con-artistry is the ability to convince people that their decisions are in their best interests when in reality they are in the con artist's. My point is that one should always look askance at another person's apparent motivations. Many of us simply take things at face value and get screwed as a result. Coercion, as such, as absolutely nothing to do with it. People that have tried to coerce me usually get the short end of the stick, but I'll admit I've been fooled a couple of times. Cost me some money too, but I took those lessons to heart and am that much harder to fool now.

Re:Enlightened Self-Interest

[rsidd]

But if someone came to me and said, "I have a mutually profitable business arrangement I would like to discuss with you" I would at least listen

Somebody called Abacha mails me about that every day. I must have received a few hundred of his mails by now. Maybe I should listen to the guy. Apparently his father was the president of Nigeria.

Re:Enlightened Self-Interest

[ScrewMaster]

Yes, I think I've received that one too, except that it was his brother-in-law.

Re:Enlightened Self-Interest

[Anonymous Coward]

Enlightened is very subjective. This still does not rule out taking advantage of someone if you think that they think that the deal is mutually beneficial, but you know it's not, and that there is very little chance your knowledge of this will ever come to light. I've yet to hear a convincing argument from an enlightened self-interester on why not to screw the other guy over in this instance, that doesn't come down to something that sounds a lot like kindness or that the overall goal is "goodwill towards men" or something like that. I happen to like dealing with enlightened self-interested people, because I find they are usually kind, but just don't want to admit it. I myself am similar in my dealings - I do always make sure I have an interest, and I often think of "kindness" as sort of a tie breaker when I have to choose between two things that give me more or less equal benefit. In personal matters, it is usually the opposite, kindness takes the forefront and self-interest is the tie breaker. A self-interested analysis of this is that I think why this works in personal matters - friends are people to whom kindness means a very similar thing as it does to you, so that being kind usually does not mean taking a loss. The thing that I always find that comes up when discussing these kinds of things with enlightened self-interesters is over the definition of 'altruism'. Many say true altruism really does not exist. And, I am of the opinion that it's their idea of what 'altruism' is that is wrong. If the definition you have for it is something that does not exist, what use is the word then?

MS Access

[Sartak]

I'm still awaiting an Access port. Then we'll have a truly fantastic Linux database. Til then, I'll stick with my trusty flatfiles.

Re:MS Access

[Anonymous Coward]

Does MS Access offer that much more then flat-files ? :)

Re:MS Access

[quanticle]

Have you tried CrossOver Office?

Re:MS Access

[DanteLysin]

lol. Wish I could mod that post as funny. :p

Re:MS Access

[caudron]

I'm still awaiting an Access port

You wanna read MS Access files in linux? Done: http://mdbtools.sourceforge.net/

You wanna port that data to an F/OSS db? Done: http://www.oracle.com/technology/pub/articles/gagn e_access.html

You want an MS Access equivolent for linux? Done: http://software.newsforge.com/software/04/04/20/18 23249.shtml?tid=150&tid=72&tid=82

Yeah, it was a joke, I know, but beleive it or not, there are those for whom MS Access is a working requirement who might be interested in these links.

Tom Caudron
http://tom.digitalelite.com/programming.html

Re:MS Access

[moreati]

I'm sure you mean this in jest. However, the folks at mdbtools are working on it.

Alex

Re:MS Access

[Nutria]

You wanna port that data to an F/OSS db? Done: http://www.oracle.com/technology/pub/articles/gagn e_access.html

Since when is Oracle a Free/Open Source database?

Re:MS Access

[LurkerXXX]

You want an MS Access equivolent for linux? Done: http://software.newsforge.com/software/04/04/20/18 23249.shtml?tid=150&tid=72&tid=82

From TFA you linked to:

"Today, Rekall is a dual-licensed GUI database front-end with aspirations of becoming Linux's answer to Microsoft Access."

" Note that Rekall does not include an RDMS -- it's only a front-end."

Done? I think not.

Re:MS Access

[SeeMyNuts!]

Well, if you are talking in terms of big database companies being very good at opening and freeing our bank accounts of that pesky burden of calculating interest on our balances...then, yes, they are free and open.

Re:MS Access

[moreati]

" Note that Rekall does not include an RDMS -- it's only a front-end."

Done? I think not.

MS Access is 'only' a frontend to JET (aka MDB). It's just a very tightly coupled frontend.

Rekall like most Unix software is designed to be loosely coupled with other components. Choose your own backend. If you don't like Rekall, might I suggest Knoda (which I believe can do a one-file-forms-n-all app), Glom (postgresql only, but tightly coupled) or OpenOffice.org Base.

Of course none of them are as ubquitous as MS Access.

Re:MS Access

[caudron]

Good point, I posted too quickly. :) All I really meant by that was that using the same basic techniques demonstrated there, you can move it to an F/OSS like PostgreSQL or MySQL.

Tom Caudron
http://tom.digitalelite.com/programming.html

Re:MS Access

[Nefarious Wheel]

MS Access is 'only' a frontend to JET (aka MDB)...

Please correct my logic here, but wasn't Access originally a rewrite of dBase IV? The same one developed by Tom Rettig, of "Lassie" fame? Is that why any Access database I've ever had to depend on barks at passing cars when it reaches 25MB?

Not entirely fair, of course -- only refers to pre-W2K MDE's. It's still only a short step from there to a decent database though, and many folks start with Access as an easy kick-start. More than that though, and the path is fraught with pain...

Re:MS Access

[Anonymous Coward]

Thanks for mentioning postgreSQL first.

Re:MS Access

[zevans]

The main problem in MS World is the number of people that use Access to do spreadsheet tasks and Excel to do relational tasks...

For small databases (ie 100,000 rows) it doesn't actually matter that both have crappy data engines, because these days Xeons eat small datasets for breakfast.

Bollocks

[teslatug]

The assertion that Linux is helping Oracle gain grounds on IBM isn't supported at all in the article. DB2 also runs under Linux, and if they said that Oracle on Linux is faster than DB2 on Linux then I could understand it. But just because Oracle runs under Linux, doesn't mean existing DB2 customers will jump ship to Oracle. Also new customers in the market for a database will not go to Oracle just because it runs under Linux because so does DB2. They'll compare the merits of the databases, and the costs that go with them.

Re:Bollocks

[Khuffie]

Yes, but new customers who already have a Linux environment running previously had one choice: DB2. Now Oracle has a chance of making sales from these customers

Re:Bollocks

[Anonymous Coward]

Oracle has been running on Linux for many years. In fact, Oracle runs on just about any OS. The biggest difference is that you can port your Oracle applications from one OS to the next with no code changes and your applications should work (you still need to test - more likeley it would be a bug in the version that you ported to that may crop up). DB2, different code base for each OS that it runs on (which I think are AIX and one or 2 Linux versions). You would most likely have to change your code to make it work to get it to work on the different OS.

That, is a big deal.

Re:Bollocks

[Slithe]

>> Yes, but new customers who already have a Linux environment running previously had one choice: DB2. Now Oracle has a chance of making sales from these customers

The Oracle Corporation released the first Linux port of their database to in August 1999, and, according to http://en.wikipedia.org/wiki/Oracle_database#Versi on_numbering_conventionsWikipedia, Oracle was the first commercial RDMBS with a Linux port. So DB2 was the latecomer.

Re:Bollocks

[leoxx]

DB2, different code base for each OS that it runs on (which I think are AIX and one or 2 Linux versions). You would most likely have to change your code to make it work to get it to work on the different OS.

Wrong. DB2 is built from a single code base on every platform it runs on. Currently that includes Linux, AIX, Solaris, HPUX, and Windows.

Re:Bollocks

[aralin]

You talk about it like this was something Oracle just did to catch up with DB2. Oracle has been on Linux before DB2. It is ported to linux since 8i version and Linux has been the primary development platform for Oracle RDBMS since the version 9i. Which means that it is developed on and for Linux and Solaris and others are now just ports of the Linux version. Oracle has been pushing the Oracle on Linux for so many years now, touting the cheap hardware to offset the cost of licenses, I wonder there is still someone who considers this as a new development.

Re:Bollocks

[ahmusch]

You're both wrong.

DB2 also runs on the mainframe -- you know, OS's like OS/390 and z/OS. That DB2 -- which is the one most people mean when referring to DB2 unless they say DB2 UDB -- is from a completely different code base, and works very, very differently. Little things like "the number of columns that can constitute a unique index" and "how partitioning works" differ wildly between the two DB2 "Universal Database" implementations. So if you want to move databases off z/OS onto a z/Linux partition or onto AIX, well, you really can't. Certainly not simply and easily. (There used to be three code lines for DB2, but they finally renamed DB2 for AS/400 to DB/400.)

However, Oracle Server has the same implementation rules -- columns per key, block sizes, how partitioning works -- on every platform where Oracle Server runs. That's the big portability difference. You can migrate a database to a different O/S and you won't have to redesign or kludge something up so it looks the same.

It is the same. That's the difference.

Re:Bollocks

[tpv]

You make it sound like there are only 3 databases in the world.

MS SQL Server is the only large scale database I know of that doesn't have a linux version. But that's hardly surprising since they don't seem to have any unix versions at all. (Hmmm... intriguing)

• http://www.sybase.com/linux
• http://www.borland.com/us/products/interbase/index .html

etc.

There are more than 2 unix databases commericially available.

Re:Bollocks

[IdleTime]

I ran 7.3.2 of SCO Unix version of oracle under Linux back in 97.

There were internal version of the database running natively under Linux long before it was released.

a bit strange

[MORTAR_COMBAT!]

since i have had much, much more success with db2 on linux than with oracle on linux. as always, YMMV...

Re:a bit strange

[Anonymous Coward]

Strange indeed!
Having come from a screaming Informix on Linux OLTP shop (Dell boxes with HP EVA storage) I wonder if IBM just forgot to pay their dues to WinterCorp last quarter.

Legally reverse-engineer Oracle now?

[Anonymous Coward]

If Oracle linked to LGPL libraries and distribute works based on it under their own EULA that is not LGPL, then their EULA (their own terms) is required to allow reverse-engineering.

If Oracle's EULA forbids reverse-engineering, then they are clearly violating the LGPL.

How do we enforce this? Contact fsf.org?

If you don't think this is true, please quote the LGPL section to support your claims before flaming.

Re:Legally reverse-engineer Oracle now?

[WindBourne]

Exactly what can not be reversed engineered? The DB? You are welcome to do so.

Postgres is heading in that way by its support of plsql and other capaibilities.

Of course, it does not have the speed of Oracle. But if you want speed for simple queries, try MySql (or even slqlite). Of course, you will give up a great deal of capabilities.

And finally, you have DB2, which competes well against Oracle on speed (lite on capabilities).

Re:Legally reverse-engineer Oracle now?

[DaHat]

It's kind of saddening when OSS zealots like yourself do not understand the very licenses that are the foundation of what you hold so dear.

Once again class... you can link against LGPL libraries and are only required to make available the modified source of the LGPL library, not the apps/libraries of yours that are using it.

Now under the GPL (notice the missing first L) that is a different story and depending on the kind of linking involved then yes, if the external libs were GPL and they linked against them in a very specific way... then they would have to release the code of the libs/apps of theirs that use it.

Guess what? By your own admission they are using LGPL libs, and no doubt any GPL libs they are using are done in such a way to permit them from having to release their core source.

Please, sit down and read the GPL and LGPL licenses one more time... and maybe, read the FSF FAQ's on both which give a better idea of what you can and cannot do with GPL and LGPL software.

Re:Legally reverse-engineer Oracle now?

[Theatetus]

Eh... honestly in all my work with databases, I've wished for a lot of things but never "I wish this database were faster". I've always wished it had more and better referential integrity options, more inheritance features, better filesystem interfaces with more frequent writes, but the problem has never been "this database isn't doing queries fast enough". There have always been some queries that were slow, but they have been from dbs that are poorly designed -- a poorly designed query into a poorly designed database will probably still run slow on a "faster" database.

Re:Legally reverse-engineer Oracle now?

[Anonymous Coward]

You obviously never read the LGPL.

Linking with LGPL library can produce 2 categories of resulting binaries:

1. derived works
2. works based on the library

Further, LGPL gives you the choice to distribute those as LGPL or "under your own terms" provided that "your own terms" meet certain criteria.

That certain criteria imposed on "your own terms" includes the requirement of allowing end-users to reverse engineer the works linked to the library.

Please. Go read the LGPL before you spout off misinformation.

Oracle Installer Sucks

[dretay]

I have recently had a lot of experience trying to install Oracle 9i on linux. The installer is broken in multiple places, and the only way to get it to install is to buy an Oracle support contract (there are specific "coyprighted" scripts that can not be found on the web). Even after you buy the contract, you have to go through several permutations of apply this patch, run this script... to get it to work. Once working the database becomes a resource hog, and seems to break quite often when I am applying system updates. Trying to move database tables from one server to another is also a major pain in the ass (although it could just be that MySQL is very easy) I can not think of anything short of a gun to the head that could convince me to try installing oracle on linux again.

Re:Oracle Installer Sucks

[IANAAC]

I can not think of anything short of a gun to the head that could convince me to try installing oracle on linux again.

I think that OTN is probably one of the better resources out there for all things Oracle (Linux or not). If you've not gone there, I highly suggest it. You'll find answers to prety much anything you'll run up against.

That aside, if your company wants to run any major commercial software on top of that database (HR, CRM, Financials, etc.) it's most likely going to have to run on top of either Oracle or DB2.

Both Oracle and IBM have staked a sizeable portion of their business on Linux. I realize some people can see the positive in that. But there is a positive in there.

Re:Oracle Installer Sucks

[dknj]

wow, first of all this is a truely misconceived post. second, shame on the moderators that are artificially increasing the validity of this.

first of all, any worthwhile company running oracle WILL have a support contract. period. second, the installer (which i agree blows chunks) is not broken to the point where you REQUIRE a support contract. i speak from experience here, because i have had no problem installing oracle 9i on linux (without calling oracle support). third, oracle is a resource hog because of its design. you don't use oracle for a 10 record database, go use MySQL and worry about your tables getting corrupted for that. now when you're talking million records or more, then oracle will "hog your resources" to ensure you get lightening quick responses while ensuring data integrity. finally, moving databases from one server to another is hardly a pain, if you are a competent system administrator (or even if you're not.. if you have a competent technical lead). again, i'm speaking from experience.

and if you still hate oracle because it is a commerical product (i say this only because you sound biased towards mysql), then jump ship to postgresql. besides the fact that its autovacuum package also sucks ass, it is the only competitor to oracle that i would trust in a production environment.

Re:Oracle Installer Sucks

[dwkunkel]

Installing Oracle on Linux is a non trivial process, but it is well documented by both Oracle and Werner Puschitz. I would recommend installing Oracle 10G-R2 rather than 9i on either CentOS 4.3 or RedHat AS4.

Re:Oracle Installer Sucks

[Anonymous Coward]

You should be using express for what you are probly doing anyway. There's a deb and an rpm version.

http://download.oracle.com/otn/linux/oracle10g/xe/ 10201/oracle-xe_10.2.0.1-1.0_i386.deb

Re:Oracle Installer Sucks

[jadavis]

now when you're talking million records or more, then oracle will "hog your resources" to ensure you get lightening quick responses while ensuring data integrity

You're implying that consuming many resources is required to ensure data integrity. It's not. Correct programming is required to ensure data integrity. "Hogging resources" (i.e. allocating lots of memory) is useful for acheiving better performance on a large database.

You also imply that a million records is a large database. It's not, unless of course each record is large. A million records can often fit entirely in RAM.

use MySQL and worry about your tables getting corrupted

Is data corruption on working hardware a real problem for MySQL?

[PostgreSQL's] autovacuum package also sucks ass

Constructive criticism, please. Autovacuum is useful to many people. If there's something about it that can be improved, let us all know.

Re:Oracle Installer Sucks

[Anonymous Coward]

Care to elaborate? I've never had a single problem deploying Oracle on Solaris and Linux. Were you running one of the supported distros or were you just trying to shoehorn Oracle onto a Debian install just because you felt it was cool? Oracle's installer works very well if you read the instructions and stick to supported hardware/software. It probably won't work on your 486 running Slackware though, but neither will DB2.

Re:Oracle Installer Sucks

[bjohnson]

Trying to move database tables from one server to another is also a major pain in the ass Hmmm: exp and imp aren't all that hard to use. You do have to answer horribly difficult questions like "Export table data? y/n" and you can move the entire tablespace in on swell foop.

Re:Oracle Installer Sucks

[idfubar]

I think the comment about using the RPM is a good idea (so is trying it with 10G or the "community edition"). I don't doubt you had problems but maybe it was a bad set of binaries (did you check the MD5 on the download)? I'm also sure that Oracle is all-too-happy to sell you a support contract but I wouldn't go so far as to say "You can't install Oracle on Linux without a support contract"; I took an Oracle class at community college and this was one of the things multiple people in the class did right at the start.

Re:Oracle Installer Sucks

[aralin]

The installer in 10g has improved radically over the previous version. You should give it a try. It has been a long outstanding problem and they really made an effort to do something about it. It will be even better in the next version from what I've seen.

Re:Oracle Installer Sucks

[ahmusch]

Heck, he could set up database links and:

1. Reference the remote table directly.
2. Set up replication to continuously replicate the tables at the remote database to the local database.
3. Use import/export (or impdp/expdp) to create a dump file which is portable cross-os and cross-release.
4 Issue "create table X as select * from X@otherdb;" ... but I'm guessing anything that might require reading the documentation will be considered way too hard for the GP.

Oracle is a big, complicated piece of software. If the GP wants toy software, well, he knows where to find it.

Re:Oracle Installer Sucks

[Anonymous Coward]

Is data corruption on working hardware a real problem for MySQL?

Yes, it is.

http://sql-info.de/mysql/gotchas.html

Re:Oracle Installer Sucks

[jadavis]

I understand that MySQL has consistency problems, due to a lack of type checking and so forth. I am not in any way endorsing MySQL.

I meant, does MySQL actually corrupt data so that MySQL itself cannot understand it? I use MySQL when I must, and I'd like to know if there are situations in which MySQL actually corrupts data.

Re:Oracle Installer Sucks

[Anonymous Coward]

I will have to say, that I don't agree, I'm running a data warehouse on RH ES3, and the installation was no problem. Installing it on fedora core is another matter. The installer can have some problems if run through VNC.

Re:Oracle Installer Sucks

[Anonymous Coward]

Okay.

I want to make and use a fresh 100,000 row temp table in postgres without:
- having to explicity analyze the table
- having to wait until the current "autovacuum naptime" is over for it to vacuum/generate statistics

I don't mind actually putting "vacuum analyze" or whatever in my application when I need to be certain the statistics are correct on a table for a very recent large change. What I'd like is to not have to think about that sort of thing. Am I dreaming?

Re:Oracle Installer Sucks

[jadavis]

There are a few issues here.

First, do you want the statistics because PostgreSQL is choosing a bad plan without them, or do you use the statistics directly?

I'm having a little trouble understanding exactly what you want. It sounds like maybe you want something other than an elapsed time to trigger autovacuum on that one temp table so that the statistics are updated. But really, if there is any time lapse at all it seems like your application could not rely on the numbers without doing an explicit "ANALYZE" or "VACUUM ANALYZE".

Perhaps what you're looking for is up-to-date statistics of some kind. This has been discussed at length on the pgsql-hackers mailing list, primarily regarding the "count(*)" aggregate function. To do it in a transactionally up-to-date way requires triggers. If you want it to be automatic, certainly triggers are not for you since it's easier to do an "ANALYZE" than create a trigger. It could also be done with extensive locking.

The main problem is that the behavior you want can't be the default because it would cause performance problems for other applications. And you also say you don't want to do it explicitly.

Can you describe what you're trying to accomplish? Perhaps there is another way.

Re:Oracle Installer Sucks

[Anonymous Coward]

Thank you for your time. You pretty much answered my question.

The real deal is that I'm trying to sell my boss on PG. He's a MySQL user. MySQL is "easy." He does reports using sales data all the time, often using temp tables with several tens of thousands of rows. If I can make the transition easier for him (ie not worrying about having to vacuum/analyze) then that's all the more better.

Re:Oracle Installer Sucks

[jadavis]

It's possible that he's using temp tables just to temporarily materialize a query result as part of a sequence of operations on the data. It's much better to use a view in that case, and PostgreSQL will optimize it to avoid uneccessary disk I/O.

And if he is using temp tables, it's still not obvious to me why they need to be vacuumed or analyzed.

Vacuum doesn't make too much sense for reporting, since there isn't any reason to have deletes/updates (at least no reason obvious to me). And analyze basically has two purposes: to allow PG to make a better plan, or if you want to use those statistics directly.

If you want the statistics to help make a better plan, that means PG needs to have several plans available. Perhaps if you're doing lots of joins, that may help. But the general use case for a temp table has just one option: sequential scan.

If you want to use the statistics directly, it seems to me better to just use aggregate functions. "ANALYZE" needs to scan the whole relation anyway, so why not use a precise query that gets exactly the information you want rather than whatever ANALYZE happened to collect?

You may have heard that PostgreSQL always needs to VACUUM/ANALYZE to be effective. That's partially true, but I think your temp tables may be a special case and not need VACUUM/ANALYZE at all.

Re:Oracle Installer Sucks

[marafa]

since i have installed oracle on linux on brand new hardware several times and plan to do so in the few future, i have made my own scripts. 2 in fact, 1 to customise the linux (rhel) environment to make it similar to the rest of the linux servers in the server farm and the other installs oracle 10g or 10gR2.
the entire process (linux + oracle) now takes 2 hours to install. i then hand over the server to the dba to continue with his work.
in conclusion, i find it pretty simple. and yes, i understand what u mean by broken, but i have fixed whatever i found that needs fixing via the scripts. as to "resource hog" u need a good oracle dba to fine tune it, consider investing in one with certification.

Re:Oracle Installer Sucks

[akadruid]

Disclaimer: I work with Oracle databases for a living and the linux box I'm typing on runs no less than six oracle databases at the moment.

However, I am having difficulty understanding your problems with Oracle 9i on Linux. I have installed 8, 9 and 10 on different flavours of linux and never once seen that level of problems. What distro were you using? Redhat, Suse, and Debian/Ubuntu on x86 are all simple installs using standard Oracle. I can't speak for the others but to be honest if you're installing on uclinux or openwrt then you deserve the pain.

Oracle on linux is easier then any other platform I have worked with. And that includes Windows, HP-UX and AIX.

Re:Oracle Installer Sucks

[Anonymous Coward]

You've posted MySQL and Oracle in one post. Bad. You've been installing Oracle - you should be sysdba. I mean SYS.DB.A, so competence is necessary to perform such a trivial task as _INSTALLATION_. Believe me or not, i've installed Oracle 9 first time few weeks ago for large hospital db app to develop module for it. Took me 3 hrs, no problems, in fact it was enouhg time to find and apply improvements to tablespace organisation. Previously I user postgres and MySQL - kiddies (PG is teen) in DB market as I see now. It's not i'm cocking,this is complex db world, but get it straight - installation shouldn't be an issue for experienced dba. DBA! BTW, db world currently suxx because of human standards - SQL92/99/etc are (_ASCII_ 70's) language vocabulary for binary machines, where is the speed when there is no way to achieve speed, without careful basic rethink of systems it would be LAMP, WIOA, LAPP OR LADA 4ever (L-linux, W-windows, I-isp, A-apache/asp, D-db2, P-postgrs/php) :/

Good news but not well interpreted

[openfrog]

It is welcomed news that Linux and open source foster a productive cooperation in the high-end database market. The interpretation given in this article gets it just backward, wrongly positioning Linux and IBM in opposite camps (facts given in the article don't support the interpretation offered). Who the # wrote this article?

Or...

[buddha42]

You could just as easily say the opposite, Oracle is helping linux.

Re:Or...

[Anonymous Coward]

I agree with this statement much moreso than I do with Linux helping Oracle. Oracle already has its stake in the ground as one of best high performance DB suites. My company recently migrated our Oracle databases from SunOS to linux. I'm sure others are doing the same. The cost of a linux server is much less than a typical Sun/Unix server, though I don't know the numbers.

RMS Access

[boomgopher]

I'm still awaiting an Windows port of Richard M. Stallman. Then we'll have a truly fanatic ex-Linux pundit. Til then, I'll stick with my trusty gefilte fish.

*yawn* sorry, tried my best, still too sleepy..

Our experience with Postgresql

[Anonymous Coward]

Oracle has good reputation for working with large size dbs. It's not cheap though, at ~$50K/CPU.

We put about 210 million records in Postgresql database for one of our apps and so far Postgresql has shown itself really well. Queries are quick, database is stable, backup times are reasonable... personally, Postgresql has exceeded my expectations.

It's good that Oracle runs on Linux, as Postgresql has done for many years, but at what point do you really need to spend all that money on Oracle? I think Postgresql will be more than sufficient for 95+% of all apps out there.

Re:Our experience with Postgresql

[briansmith]

It's good that Oracle runs on Linux, as Postgresql has done for many years, but at what point do you really need to spend all that money on Oracle? I think Postgresql will be more than sufficient for 95+% of all apps out there.

I agree, but I would like to point out that Oracle doesn't usually cost $50K/CPU for any system that would be sufficient for PostgreSQL. It is more fair to compare Oracle Standard Edition or Oracle SE One to PostgreSQL, which are priced significantly lower ($15K and $5K respectively, plus support). Even EE is "only" $40K/CPU, plus support.

Re:Our experience with Postgresql

[CastrTroy]

$5K plus support? wow, that sucks. For $5000 they'd better throw in some support for free. And all this Per CPU stuff really gets on my nerves. Oh, sorry, you have good hardware, so you have to pay more. No wait, you don't have good hardware, you have an old dual CPU Pentium Pro, but you still have to pay more. If they want to charge more for people who have more power, they really should charge per cycle.

Re:Our experience with Postgresql

[MoogMan]

Large companies (Enterprises, maybe telco carriers) will need to have a decent support contract, as well as guaranteed uptimes with clustering and all that shizzle. This, I think, is why Oracle is bought.

Sure, Postgre has support, but I'm not entirely sure that when I email the outsourced support company, or ring their phone number, that someone will pick up. It's this uncertainty that makes the higher management simply decide to go with Oracle/DB2/etc.

Forgive my ignorance, but I do not know if Postresql supports multi-system clustering, so I will refrain from commenting about it.

Re:Our experience with Postgresql

[jadavis]

If they want to charge more for people who have more power, they really should charge per cycle.

But we all know that clock speed is not an accurate measure of processor speed. What they should really do is tie the pricing to benchmark results on Tom's Hardware.

Re:Our experience with Postgresql

[jcaren]

"Sure, Postgre has support, but I'm not entirely sure that when I email the outsourced support company, or ring their phone number, that someone will pick up."

As Nelson Munce says "Ha!!!". From recent experience, Oracle under linux sucks big time! If you use MTS it crashes ~every 24hrs. The fix from oracle - reboot at midnight :-)

As for commercial support Oracle seems to be staffed by the same hopeless indian script readers that everyone else is using. I certainly would not consider paying money for that sort of "help".

OTOH open (read free) PostgreSQL is well supported and commercial support is highly recommended and rated by people I know that have used it - no indians reading from autocues!

The problem - oracle is no longer a database company. It became an enterprise application vendor around oracle 8 and ever since then it has modified functionality with the objective of eliminating "competing" suppliers - who have since moved to other database vendors.

Now it wants to recapture the DB market but still has its "app vendor" hat on and treats other apps providers with outright hostility.
Until they get thier DB vendor hat on (esp.y on support) they will continue to rate below Both Pg and MySQL - even if they "own" core MySQL technology now :-)

P.s. Although I am a long time Oracle developer (pre 7.0), I am doing quite a bit of Oracle to Pg migration work these days, so thanks to Oracle for driving my customers to me :-)

Re:Our experience with Postgresql

[briansmith]

$5K plus support? wow, that sucks. For $5000 they'd better throw in some support for free. And all this Per CPU stuff really gets on my nerves. Oh, sorry, you have good hardware, so you have to pay more. No wait, you don't have good hardware, you have an old dual CPU Pentium Pro, but you still have to pay more. If they want to charge more for people who have more power, they really should charge per cycle.

They used to charge Mhz * CPU's * platform-rate, but people thought that pricing was too complicated (and, I kind of agree).

I mostly agree with your sentiment: if PostgreSQL can do it, then use it instead of Oracle. But, in most cases, if Oracle has a feature you could signifantly benefit from, and PostgreSQL doesn't, then it is probably going to be cheaper to buy it from Oracle.

Also, don't discount the value of Oracle Support. I've never had better support than I got from Oracle in the two cases where we had critical problems. Oracle support has 24x7 30 minute response time and will spend hour helping you when you have a critical problem. In order to deploy PostgreSQL instead of Oracle, you have to take several measures to ensure that you will never, ever have any problems where this level of support is necessary.

Re:Our experience with Postgresql

[CastrTroy]

if Oracle has a feature you could signifantly benefit from, and PostgreSQL doesn't, then it is probably going to be cheaper to buy it from Oracle

Actually, if you need the feature, then it may be cheaper to pay someone to implement the feature in Postgres. Unless there's some patent or other reason why the feature can't be implemented in Postgres, It's probably better to have it implemented in Postgres. It may not be cheaper for 1 company, but probably for all the companies that need the feature, especially once you factor in the cost of future upgrades.

Re:Our experience with Postgresql

[Anonymous Coward]

Try implementing this oracle hack for your boss income in postgres, you'll see practical reasons for reasoning: SELECT * FROM (SELECT * FROM salary ORDER BY income DESC) WHERE ROWNUM=1

Out of curiosity ...

[vlad_petric]

What's the main selling point of Oracle to begin with? From what I've seen so far, most IT people dislike it (pain to install on Linux, huge memory requirements, optimizer that needs many hints, etc).

Re:Out of curiosity ...

[mdfst13]

1. Oracle has good (albeit expensive) support

2. Oracle is very customizable

3. Oracle is very powerful (in terms of expressiveness of its custom SQLish statements)

4. Oracle performance is incredible when properly tuned

That said, Oracle is not a newbie database. It won't configure itself for you like MS SQL does. It expects to be operated by a professional who does nothing other than work with Oracle all day.

In particular, "optimizer that needs many hints" is a sign of the power of Oracle. They assume that if you are optimizing, that you want to wring out every last bit of performance. Thus, they give many options but require you to take the time to analyze your data and tailor the optimizations to your data. Oracle rarely trades power for ease of use.

Oracle's not designed to be a push button solution. It's designed to allow expert users to do everything they really want to do.

Re:Out of curiosity ...

[jadavis]

In particular, "optimizer that needs many hints" is a sign of the power of Oracle.

It's also a weakness. The plans should not be static because your data is not static. That means that when the data set changes, you need to re-optimize.

Re:Out of curiosity ...

[jbolden]

Oracle 10 (I can address different versions if you like) uses a very complex strategy to determine what the best sequence is for a SQL computation:

1) Automatically generate statistics as you use tables
2) If those statistics are good enough then using the statistics estimate costs for various strategies
3) If those statistics are not good enough then use a rule based system

It used to be that hints were used for rule based. However there are used for times where statistics are not likely to reflect the query. There are examples given but its basically an engine override. Hopefully you know what you are doing when you override the statistical data.

Re:Out of curiosity ...

[greenechidna]

I dn't know what version of Oracle the people you talk to have been using but I find that the optimizer works very well without hints. The thing I like about it is the power of the PL/SQL language and its support for complex datatypes. I also like the use of packages to group procedures and functions together logically. I can't say how this compares to PostgreSQL as I haven't really used any open source databases as the organizations I work for have, in the past, been a bit suspicious of Open Source. This is changing a little as we now have quite a few Red Hat boxes and are in the process of deploying Oracle 10g (RAC) on Red Hat.

Re:Out of curiosity ...

[ozsynergy]

I've been working with it solidly recently and have to say it has some good points. - It has some really neat scalability using its Grid technology. All you need to do is swip the trusty gold credit card. - It has got XML tables which are very powerful and allow you to use the power of sql to query xml without extracting all the data out. - Theres inbuilt apache with php support pre-built - HTMLDB now known as Oracle Application Express is a free bonus excellent for building in house stuff and maintance systems in very little time. - Integrated J2EE support There are lots of others good things about oracle It has alot to offer. I recommend it as far as the Database server and the companion (apache/php/oracle application express) All the 'infrastructure services stuff' may be good when oracle them selves understand it :)

linux helping oracle? Other way around?

[pedantic bore]

Wait: Oracle is porting their database to Linux and the headline is "Linux Helping Oracle?" Sounds more like Oracle is helping Linux get into the datacenter, or at the very least, there's a symbiotic relationship. Linux is useful to Oracle (this isn't news, they've been pushing to an all-Linux solution for years) but Oracle is also very useful to Linux. Hate Oracle if you must, but admit that they've put a lot of money into Linux.

It happend this year all of a sudden?

[Anonymous Coward]

So this company WinterCorp just now notices that there are open source databases in use out there?

Hmm.

My money is on IBM

[Anonymous Coward]

Eventually, open source databases will be as capable as Oracle. Then where will Oracle be? IBM, on the other hand, seems to understand how to make money in an open source environment.

The increasing capabilities of open source databases will drive Oracle to retreat up market until there is nowhere left to retreat. Oracle is not in a position to stand and fight (although it seems to be working on the principle that if you hire all the database developers, you can stop open source database development). Oracle will have to re-invent itself. That process is never guaranteed to be successful.

Not too surprising

[iabervon]

I've used Oracle on both Linux and Solaris. We were running it on Linux workstations, along with a bunch of other things, for development, and on Solaris to test it in the conditions we expected it to use in production, with nothing else running on the database server. Even so, we found that it was faster in the Linux setup. Of course, this is a while ago, and on relatively small data sets, and not an especially high-end Solaris machine, but it was still striking. At the time, at least, if you could get a big enough Linux box to handle your data, it would cost probably 1/20 of the cost of a Solaris system capable of the same performance. You'd probably have issues if you had to move to a database cluster, because the software for getting a cluster to work wasn't so good (when we tried it a few years later), but for a range of useful sizes, Linux was a much better value than Solaris.

The issue with clusters seemed to be that it was only available in a special custom Red Hat configuration, and it wasn't well tested, because it was just Red Hat and Oracle doing it, not everybody looking over the patches for whether they would screw up the rest of the system. There's been a bunch of merging work on cluster stuff since then, so the code quality is probably now up to the usual standards.

(Of course, I've been just as happy with other database programs on Linux, but I don't have any experience trying to use them for insanely large or busy databases; our program was supposed to work with huge data, but we never used it on huge data ourselves.)

Linux cluster file system

[jbolden]

Wow between a bad CNN article and a bad editor this article makes no sense at all. OK Oracle has been on Linux for years. The CNN article that got linked to is about Oracle porting their cluster file system over to Linux. This is a filesystem that is just a little faster because it removes some redundancy. To quote Oracle:

Cluster Filesystem Options for Running Oracle
Oracle RAC technology already provides features such as load balancing, redundancy, failover, scalability, caching, and locking, so there is a duplication of functions that occurs when Oracle datafiles reside on a block device with a traditional Linux filesystem such as ext2/ext3. Performance decreases in this case because caching by Oracle as well as the filesystem drains memory resources.
As of this writing, in addition to third-party cluster filesystems, there are four filesystem options for running Oracle RAC. They are, in order of recommendation by Oracle:
 
1) Oracle Automatic Storage Management
2) Oracle Cluster File System
3) Network Filesystem
4) Raw devices.

Surprise, surprise

[Britz]

Oracle didn't port their database to Linux for charity or because they love open source. They went in for the money. What a surprise. They wouldn't have known Linux existed if they wouldn't have been requested by customers, because with i386 you have been getting more bang for the buch for some time now. And Linux is THE i386 Unix. I am talking big corporate supported stuff, I know that BSD is cool and all. But Red Hat on Intel servers has been a very good deal in many cases. And if the customer doesn't spend as much on hardware they might be more willing to shell out the fortune that Oracle demands.

And this is news???

OracLinux

[Doc Ruby]

What's cool about porting Oracle to Linux is that Oracle can modify Linux. They can drop parts of the kernel that don't help Oracle run, and add parts Oracle needs but that isn't part of Oracle. I'd love to see an Oracle Linux distro that is stripped to do nothing but run an Oracle server (not even run Oracle clients) and maybe one of Oracle's Java app servers, in clusters.

Re:OracLinux

[idfubar]

I agree; what's even more likely is that you'll see an Oracle "Virtual Appliance" which is basically an Oracle installation on Linux/Solaris/AIX inside a VMware virtual machine. This means you simply download the VM files and hit "play" and you have an optimized OS + (reasonably) optimized DB; just create your tables, indexes, and go...

Re:OracLinux

[iamdrscience]

Are you insane? The overhead from running it as a VMWare virtual machine would far outweigh any optimizations that could be made. Plus, you have to pay for a VMWare license on top of your Oracle license.

Re:OracLinux

[ahmusch]

However, for non-production systems -- generally dev, test, and qa environments -- the use of VMware for both Windows and Linux is hot-hot-hot. Those platforms are generally woefully underutilized, where the utilization numbers are on the order of 5-10% during the working day, lower during off hours.

Collapsing several system's environments like that into a single VMware server can save lots and lots of money paid for :

Systems
Data Center Space
Data Center Operating Costs

Besides, I remember having seen something about all the VMWare releases except for the top-of-the-line one being more or less free now.

Re:OracLinux

[idfubar]

While accurate ~5 years ago your comments, in today's terms, are totally off.

First, the overhead is minimal (~5%). Virtualization is everywhere, *especially* in backend infrastructure and business applications. The flexibility that the virtual machine gives you FAR outweighs any small performance penalty. Go take a look at the list of reference customers VMware has of people who are very happy with virtualizing; many of them run Oracle databases in VM.

Second, not *all* VMware products require a license. The VMware Player is free and so is the VMware Server; what you do have to pay for is if you decide to use ESX (no OS overhead) or VMotion (to handle multiple ESX nodes).

In terms of startup time there's no "overhead" for a virtual machine; a slimmed down OS is a slimmed down OS. The same is basically true for memory footprint as well. The real savings is in terms of installation time and configuration of settings, which is obviously what the OP had trouble with.

Re:OracLinux

[Anonymous Coward]

Oracle tried this way back in the 8.x days (I think). I know HP had a "certified Oracle appliance", can't remember if Sun or IBM had such a config. Never really took off and just faded away.

As someone who makes these decisions ...

[Chitlenz]

I'd say they're right, but also this article is a tad late to the party. This has been going on for at LEAST 5 years, since 8.0 was first released for, I believe, Redhat 7. Consequently, this is not some huge rush for Redhat, and I actually have found tighter distros to run 10g better (I like gentoo, but it's a pain in the ass to get tuned right for this particular task). Anyway, what I found interesting is that our linux oracle systems absolutely STOMPED the 8 way v880/16GB Solaris boxes in archive testing involving 4+TB databases (this to us was a real shock btw... I'm currently buying v40z class servers from Sun that are 4x dual core opteron boxes for like a 10th of the price of a true solaris (Sparc) platform. Thus I would say IBM's problem is Sun's problem in this case as far as selling big iron anymore).

I think Oracle is winning because Oracle is honest to god better than their competition. I was (am?) a DBA for 10 years on Sybase (AIX), SQL Server 6x 7x 2kx, Informix 8x 9x, and Oracle 8x 9x 10x at various times, and though I've moved on to a database architecture role with the company I'm with, I'm still making the call on systems purchases. We use mostly SQL Server 2005, for cost, in the smaller 4-6TB systems and they run great, but I wouldn't even consider DB2 for any production role anymore with Oracle out there making it happen in so many better ways.

I'm not a fanboy of Ellison, I'm just realistic about who's driving the market today.

--chitlenz

PS - Oh yeah, as mentioned we're running Sun 40z's with Windows Server 2003 and SQL Server 2005 on Netapp arrays AND it is VERY MUCH worth noting that the lower end Sun/Opteron line not only runs windows, but runs windows VERY well (driver support for their servers is very very good, which was like ... well weird... 'Sun support? Can I get a download link to your windows drivers?'). Try it and be shocked ....just a tip.

--chitlenz

PPS - for anyone who is curious about this topic in any real way, use an isntall guide other than Oracle's, since it's usually wrong for awhile ... use something like http://www.puschitz.com/InstallingOracle10g.shtml
instead.

Most "IT" people...

[Anonymous Coward]

From what I've seen so far, most IT people dislike it

The kind of "IT people" of whom you are speaking, are far too untrained and unqualified to be making a valid judgement call on the merits of Oracle. Oracle is an incredibly sophisticated database system, intended to be installed and operated only by those persons with enough training and understanding of its architecture. It is intended for really big, really complex applications and not for the mundane. In its intended applications, Oracle is powerful, fast and unbeatable. It scales to levels that MS SQL and other lesser databases can never reach (DB2 is its closest peer, Informix once was too, but that's toast now). It has a steep learning curve that you must make a serious commitment to mastering, and once you've reached that expert level, you'll easily see that Oracle is the "king daddy paw-paw" of all RDBMS's.

In parallel to your statement, we could also say that "From what I've seen so far, most PC users dislike Linux" because it too has a learning curve to it that is radically dissimilar to Windows from an average PC user's perspective.

Re:Most "IT" people...

[Anonymous Coward]

Oracle is powerful, fast and unbeatable. It scales to levels that MS SQL and other lesser databases can never reach (DB2 is its closest peer

WTF? Don't you mean Oracle is DB2's closest peer when it comes to scalability and performance?

Re:linux helping oracle? Other way around?

[TheRaven64]

If you don't count Linux, then the platforms on which you can run Oracle on are: Windows Scales up to 32 CPUs, and gets really expensive above about four (although not in comparison with a 4+ CPU license for Oracle). Is not well known for security or stability, and does not have a strong reputation as a database hosting platform. Proprietary UNIX Generally is only supported on the manufacturer's own (very, very expensive) hardware. Adds vendor lock in, and cost. By running on Linux, Oracle lowers the barrier to entry for a new customer (and means that a greater proportion of the TCO for an Oracle system winds up in Oracle's pockets), so it's hardly surprising that they benefit from supporting it.

Re:linux helping oracle? Other way around?

[Anonymous Coward]

You are right, by encouraging customers to run linux, Oracle gets a bigger bite out of the IT dollar pie. Does anyone think Oracle likes sharing that $4million project budget with Sun or IBM? No, what they want is to push customers to the cheapeast hardware and the cheapest (seemingly) OS so then they can come in and gobble up a bigger chunk of that $4 million with things like extra licenses, RAC upgrades and consulting.

But, when push comes to shove, does Oracle run their business on many small Intel boxes running Linux like they encourage all their customers to do? No, Oracle runs on 4 big iron Sun boxes. That's where Larry keeps track of *his* money.

So both installer suck?

[Anonymous Coward]

I haven't bothered trying the Linux installer. I did however try to install it on a win 2k3 server 2 weeks ago... And it's the most weird/troublesome DB install I've ever done. It was just a 3GHz P4 w/ 1.5GB RAM (small server from a workstation, just for some dev work and testing - nothing production/large scale or anything). The box has always been working flawlessly (clean from a ghost image too). Nothing weird in the event log, no hardware conflicts, no old drivers, etc. And the freaking Oracle installer choked the CPU hard for 10+ minutes. All you could tell is, CPU load from the installer is basically 100%. No signs of progress or anything. Eventually I assumed it was crashed and "end tasked" on it. Reghosted (don't want to reinstall over a crashed install). Same thing the other time, except I walked away from it to go get coffees downtown. When I came back (at least 20 mins later) then it was done. (and no, it wasn't installing off an old 2x CD-Rom drive either)

Hogging 100% of the CPU for at least 15 minutes and showing no signs of progress makes it by far the worst installer of ANY software I've tried lately. (No "this will take ages" warning either)

And that was a rather minimalistic install (not even the sample DBs) from the first disc only. I can't imagine installing all 4! And right after installing (again, no data at all, barebones install, etc), not having made a single ODBC connection, without having even touched the admin tools, etc, the thing was using like 400MB of RAM. That's a LOT of overhead for a DB server that doesn't even have a DB created on it already. I can run several DBs easily in that much RAM (yes, one can tweak it to use more RAM for faster queries when needed - but 400MB can't be a copy of the data in memory - there just wasn't any data to cache yet).

It might scale well and be powerful and all, but man. You wanna use it? Make sure you got some high-end hardware. You'll need it. We've more or less given up on it. The demand for our apps (the usual N-Tier business apps) to connect to Oracle was quite low to start with, and since it's being a pain (semingly needs highly paid Sr DBAs and PL/SQL devs too)... We'll just keep using the others instead (MS SQL, PostgreSQL, MySQL, Sybase, DB2, etc). Most of our customers use SQL Server but more and more are moving to MySQL/PostgreSQL lately.

Anyways. Seeing how the installers suck, how heavy this bloated POS is, how expensive it is licensing wise (more than comparable editions of DB2 and SQL Server with similar features) and support wise (Sr Oracle DBAs don't come cheap and it seems maintenance heavy), etc, I can see why other databases are on the rise in a lot of places (often free/OSS ones). Some places will stick to Oracle as they really need it or have too much money, but that's about it.

Re:So both installer suck?

[IdleTime]

Very interesting issue, since I have never seen a similar install issue and I have worked with installation issues for over 10 years.

The problem could have easily been diagnosed if you had used a real OS and not a toy OS like Windows. I take it that you checked which process were hogging CPU and since it was the installer, I'm 100% sure that your problem was with the version of JRE installed on your machine. It would have been nice to have an RDA (Remote Diagnostic Agent) output from your machine since the problem is on your side and not on Oracle's.

What is more interesting is that you show a total lack of knowledge about oracle. 400MB is easy to reduce on a single user database, but yes this is a much more resource needy system than your average editor or browser, but then again, Oracle database is probably the most complex piece of software available on the market. So yes, it will use a lot of resources and it is not intended as a single user database. Use Express or some other slimmed down database or learn how to configure it, take a 5 day DBA class. Without it, you can just as well deinstall it or hire someone who knows it.

Personally, I have worked with all versions since 4 and we are now on 10.2.0.2, that means I have about 20 years experience with Oracle.

Re:So both installer suck?

[Anonymous Coward]

That was latest JRE at the time 10g R2 came out (or the day after), straight from Sun (attempted installs on other PCs went pretty much like that one too, took forever as well, installer pegging the CPU to 100% for half of eternity - you'd be better off compiling gentoo).

And Windows, that "toy-OS" works great for us and that's more or less all our customers use too. Saying things like that only make you look like stupid troll and linux fanboy. Works great, the only issues I've had with 2003 so far is installing Oracle (the other DBs don't have such issues either).

Anyhow. I never said I expected it to use up only as much RAM as a average user process (like browser or whatever), but by default it still uses 400MB+, which is more than the default installs of SQL Server Enterprise Ed (the whole/full install; PLUS the express ed AND mobile ed too) + MySQL + PostgreSQL together (that's 5 DBs). And Oracle is the only one which was such a barebones install. So what you're saying is, me expecting Oracle not using over twice as much memory as 5 other leading DBs makes me a n00b? Okie (nevermind I could teach that course on the other DBs...) I've seen other DBs use way more memory, but that was a production server (or perhaps apps like Exchange which also use a lot of RAM). And at that time, there was no express edition.

Oracle database is probably the most complex piece of software available on the market

You've misspelled bloated. And yes, it surely is!

Re:So both installer suck?

[IdleTime]

Well, the installation documentation states a specific version of JRE and that is NOT the latest version.

The leazt you could have done, was to follow the installation prerequisites in the installation guide. When you don't, you do get problems.

Toy-OS - Windows... Well, why don't you tell me how to trace a process under Windows with a tool like strace or truss or glance? Which tool do I use to list all open files under Windows? How can I trace a certain thread under Windows?
The problem is that Windows has zero tools useable for diagnostic of problems and for tracing. It is the most horrible OS to use and the wast majority oif stupid issues in regards to Oracle originates in the Windows world. The difference between support issues on UNIX and Windows is amazing. Many issues under Windows is due to people who think installing Oracle is like installing any other program and that you don't need to read anything nor do you have to follow instructions. This is the biggest problem with Windows, i.e the people who use it.

During the creation of a database, you have the opportunity (in the installer really) to configure the SGA. The default value is meant to be good enough for a starter database with several users and not for a single desktop PC. That is not the target and if you don't know how to configure an Oracle database, then you should really get some training because it is wastly more complex than antyhing else you have touched. If you just laugh ar this, you will have problems with an Oracle installation. It does require quite a bit of knowledge in order to work with it. And yes, it requires a lot more resources than anything else you have used too, the documentation clearly states what the minimum requirements are for a server running oracle, but again, you seem to ignore these facts.

Sure

[A nonymous Coward]

It's hard to screw up flat files enough to justify charging $$$/hour to fix things up while blaming someone else for their shitty software.

Oracle has run on Linux for ages

[James Youngman]

Yes, but new customers who already have a Linux environment running previously had one choice: DB2.

But Oracle has run on Linux for a long time - I remember installing 8.15 (well, 8.x for some value of x), and the current Oracle release is, what, 10g?

Re:Oracle has run on Linux for ages

[outsider007]

Anyone who has ever tried to install 10g on linux knows it is easier for a camel to pass through the eye of a needle.

ugh, all three wrong

[kpharmer]

Saying that DB2 has three lines of code:
      - unix/linux/windows
      - mainframe
      - as/400

while oracle only has one for:
      - unix/linux/windows

is a nonsensical comparison: Oracle doens't have any product on the as/400, and their product for the os/390 (mainframe) is practically non-existent. A more reasonable statement is:
    "db2 and oracle each have just one codebase for the common distributed platforms"
db2 has a slightly different codebase for platforms that oracle doesn't support anyway.

Next point: migration of db2 code from mainframe to/from linux/unix/windows isn't necessarily a big deal. Sure, the file systems are different, memory model is a little different, and partitioning is different. However, some of those differences are unavoidable - the mainframe simply works differently than linux or windows (doesn't have concept of directories & files, etc). And 99% of all features and skills are the same.

My team just picked up a db2 dba who's work is primarily on the mainframe. We're not at all concerned that she doesn't have aix or linux experience - the small differences are quick to learn. Of course, she is going to have to learn linux & aix in order to run jobs, view logs, etc, etc but that's not a database issue.

Re:ugh, all three wrong

[IdleTime]

10gR1 is available on OS/390 and works very well and include all the same features as any other platform.

OS/400 is a different beast and very different from z/OS (OS/390) and UNIX/LINUX world and currently this market is very well covered by IBM and most AS/400 customers are fanatical IBM customers. Not a market that Oracle put much interest in, nor is there much money to be had in this market anymore.

Oracle's code consists of 2 parts, one general part which is identical on all supported platforms and one machine dependent part which is handled by each porting group once the general release is done on Linux. Depending on the platform, this can be a very daunting task with the result of later releases on certain platforms. It all depends on which tier your platform belongs to.

Re:ugh, all three wrong

[ahmusch]

And as someone who's ported commercial software between DB2 z/OS, DB2 LUW, and Oracle, I can tell you that the "slightly different" code is not so "slight" when you're actually elbows deep in it. The devil is in the details, and that 1% you so blithely dismiss can chomp down quite firmly on your backside when you least expect it.

Oracle Server is Oracle Server regardless of where you implement it -- Windows, Unix, or OS/390. I agree that there are differences in the kernel code and the process model (used on Unix) versus the thread model (used on Windows). However, those are not differences that are relevant to the users -- the same initialization parameters and SQL syntax apply regardless of platform.

DB2 is not DB2 regardless of where you implement it.

You can migrate an Oracle database from z/OS to z/Linux with import/export. You cannot migrate a DB2 database from z/OS to z/Linux without possibly having to redesign and reimplement various schema objects.

That's a significant difference, and it's one that causes lots of heartburn for those of us who maintain applications that support multiple databases. You figure you'd only need two database code lines -- one for Oracle, one for DB2. However, you need three -- one for Oracle, one for DB2, and one for DB2 UDB.

Re:ugh, all three wrong

[kpharmer]

> The devil is in the details, and that 1% you so blithely dismiss can chomp down quite firmly on your backside
> when you least expect it.

Sure, if you're using partitioning then there will be some major differences. Of course, most os/390 db2 databases in my experience don't. The other major difference is that new features are released much earlier on LUW than on OS/390 - so if you coded for MDC, MQT, etc on DB2 you may have to wait to get it on 0S/390. And of course, given the nature of how upgrades happen on a mainframe - even once it's available it may be another 1-2 years before you see the mainframe db2 version get updated to support it.

> You can migrate an Oracle database from z/OS to z/Linux with import/export. You cannot migrate a DB2
> database from z/OS to z/Linux without possibly having to redesign and reimplement various schema objects.

Right, if you're the one guy in the world running oracle on a mainframe then the migration easy. If you're one of thousands running db2 on a mainframe then it is usually easy assuming that you're not using partitioning or building database management software that works at a low-level of the database. If you are running a major data warehouse on the mainframe or are a tools developer like BMC or Quest then yeah, you'll have a lot of work to do.

Re:ugh, all three wrong

[ahmusch]

I'll just say this much:

If you're running a medium to large system on the mainframe, it's very likely that you do use partitioning. Otherwise purging data, backing up data, or searching through billion-row indexes is not going to be as performant as possible.

Performance is often achieved by writing to the low-level. Generic code and generic SQL performs... generically.

article is mistaken

[Anonymous Coward]

This article must be wrong. Microsoft says that linux is for print and web servers only, and never, ever makes it into the data center. And we know anything Microsoft says is true.

If Slashdot had editors...

[afabbro]

...they would have included a link to the WinterCorp report.

So what about MySQL?

[Anonymous Coward]

Is there something wrong with my SQL?

I've been using it on my redhat server for more than 5 years now with lots of hits every month (over 100,000) and it performs nicely and without problems.

Why do I need some greedy company with a bloated application that just decided to show up to the game?

I think MySQL fits most any database application.

Re:So what about MySQL?

[zevans]

Nothing wrong with what you say, but you dropped some zeroes. FTSE100/Fortune500 companies run sites with that many hits PER HOUR, not month.

Maybe MySQL is good for big applications, for some values of big...but not 200-blade Linux farm big.

Reading Comprehension?

[merky1]

Is it just me, or did the article and the submission vary greatly in mission. According to the article, the only thing Oracle is doing is contributing a new file system to the Linux community. According to the write up, it would seem Oracle is porting its database to Linux, which I believe it did like 5 or 6 years ago. Now the article seems to make some wand waving to this being a competitive move by Oracle, but nothing in the article points to that.

training biases

[Joseph_Daniel_Zukige]

There are systems so complex that merely becoming competent in them strips one of objectivity. Near as I can tell, Oracle's databases are such. db2 definitely used to be, I'm guessing it still is.

That kind of complexity is useful for a while, but then the advances of Technology tend to bury the product. Some people suspect Oracle of trying to pervert sleepy cat and others by burying them, but the announcements make me think they've realized the same thing as IBM -- the money in no longer in the software, but in the services. (Something tells me there is something dangerous about that situation, as well.)

Small is the way of the future. Big institutions are only useful to the extent that they enable individuals.

Big Database

[AgNO3]

It was my understanding that one of the largest databases on the planet was Boeing's. It tracks every repair ever made to every plane. Like down to replacing a single screw. I don't remember when it was started but my step mother was a cobalt programmer for it. I remember her getting her portion y2k compliant then quiting and making some really sick money as a freelance cobalt programmer. Anyway this was a bit back but She told me that the database was so huge that No one had been able to port it to anything. Boeing was offering some huge for the time amount of money as a reward on top of the contract they would provide for any company that could get the DB out of its cobalt world. No one had been able to do it. That list from wintercorp did not even list Boeing. Is it possible there are bigger DB;s out there that are still running on proprietary apps on main frames???? ( I am sure Boeing by now has moved away from the cobalt based DBs)

Re:Big Database

[Anonymous Coward]

First of all, it's 'Cobol', not cobalt, which is a metal. Secondly yes, there's IMS on IBM 390 mainframes. Here's a free, brief, flash-based tutorial.

Oracle on Linux?

[TheNetAvenger]

Oracle on Linux?

Doesn't Linux have enough problems getting a foot in the door?

Partially kidding, but Oracle? Who in the Linux community wants to see Oracle running on Linux?

Oracle use to be quite grand but never evolved past the 'usability' model, mainly because they made so much money off of selling training. Virtually killing it for serious developers when interface and application independance became the norm for databases in the early 1990s.

Although they did learn to some degree and focused on the database technologies instead of trying to keep users locked into their usability patterns.

There are SO many database technologies I would choose on ANY scale project before Oracle. I guess I am a bit biased, but I also was trained and developed on Oracle for several years in the early 90s, and have been smacked around trying to make current Oracle technologies work in environments when IBM or MS and even MYSQL do things so must faster and easier.

I just don't see this as a big thing, except for the people that have bought the Oracle kool-aid and don't realize there are better solutions out there.

Re:Oracle on Linux?

[ces]

Um, like FOSS databases or not I just don't see them being ready for the kind of applications the ones listed in the Wintercorp report are likely being put to.

Still considering some of the databases listed were MS SQL I suppose anything is possible.

Re:Oracle on Linux?

[Stu Charlton]

I compete with Oracle daily , working for another software vendor, but I wholeheartedly recommend their database to almost any scale solution. It's a great piece of technology. And most large organizations that use Linux use it to run Oracle, or an application server of some sort, giving Oracle the largest Linux databsae market share (by revenue).

SQL Server 2005, by my estimation, is also good, but very new.

DB2 UDB has some positive traits, particularly the parallel edition for large data warehouses, but otherwise has aspects that I don't like (particularly their locking / concurrency management). Ditto for Sybase, adding in (what I feel to be) a very difficult recovery process compared to Oracle, which has some fairly well trodden paths to recovery (especially in 10g).

Re:Oracle on Linux?

[TheNetAvenger]

Don't get me wrong there are a few instances where Oracle still fits the model.

However these are disappearing because of Oracles relunctance over the years to adapt.

Oracle was more than a Database in the early years, Oracle, like dBase and other low end tools all had a 'user interface' model that locked all applicaitons using the products to be produced 'inside' the Oracle technologies.

Sure this exists less today, but Oracle's business model hasn't really 'got' the UI migration move of the industry.

Even old MS Access and early MS SQL both had development interfaces, but where the products shined is that you could use ANY tool to create an interface to these database technologies. (Using MS examples here because of the timeline and people's familiarity)

Oracle didn't 'get this', instead to use Oracle, you had to NOT ONLY develop your application in the Oracle interface, but were locked to the Oracle keystrokes for operations, doing searches, etc. Which was VERY non user friendly.

Oracle didn't really care, as Oracle did (and still does) made good money on companies not only buying the product, getting developers trained to build a solution in Oracle, and then sending all the employees to learn how to use the applicaiton in Oracle with the standard Oracle keystrokes, and modes of search the data.

This is why you found low end products filling needs that once Oracle and dBase were the answers for (having a similar UI model and development environment), and people were moving to Access and VB to Access development projects.

As you were purely working with data via the interface and SQL, and could create ANY interface construct possible.

This is what bothers me with Oracle, even 10 years later, they STILL don't full get that we don't want to WORK in their database, we want it to do its job and we will create custom interfaces in C++ to PHP Web interfaces.

(Yes you can NOW do this with Oracle, but this is not their bread and butter, and they would rather lock you into their interface and development.)

Also the costs of Oracle are incredible, it makes MS-SQL look cheap, and yet performance wise DB2, and MS-SQL outperform Oracle on almost server/client configuration. So why pay big bucks to Oracle, when you can ge a solution that has more options of usage, and runs faster, and is cheaper. (Especially when you look at their processor licensing, etc.)

Even MYSQL fills the needs of places I see companies use Oracle just becuase it is what they have 'always' used, not because it is the best solution.

We work with companies that are still in the Oracle model, and don't plan on changing because of technical legacy reasons or old IT Oracle admins that don't want to learn how to manage other solutions.

And this is really sad. I see these companies have to schedule long down times for backups, and other Database maintenance routines that are just NOT something you find with DB2, MSSQL, or even MYSQL. The next time you visit a major companies site, and they have a daily downtime policy (like a bank, or even Sony's Online servers), it is because they are using Oracle. Sad...

And even more important...

[emil]

...is the fact that the only verison of DB2 that is fully instrumented is on the mainframe - it is in the end impossible to fully quantify performance problems under the Windows/UNIX/AS400 platforms. Oracle is fully instrumented everywhere.

Of course, I read this in some Oak Table literature, so I wonder if I should trust it fully.

Article Translated

[sihker]

Steps to install for known partners(RedHat,Suse) etc. will be shortened from 20 steps to 15. Glibc strict requirement will have to wait until next big announcement.

The Oracle Cavorticle

[GNMorley]

After reading all the posts, there doesn't seem to be an Oracle customer in the house. The clueless leading the blind. Gee, all of the Oracle databases are on Linux? Duh. If you've been part of the process in the last 5 years you know that Oracle pushed all it customers to use Linux. If you are investing 7+ figures in a new system, and Oracle favors Linux, how many shops are going to go against Oracle's not so subtle advice? How many shops are going to run Solaris or Windows if Oracle and it's partners push linux? Is it because Linux was a better solution? Linux FANBOIs, keep your pants on. Why Linux? Because Larry hates Bill, is very jealous of all that cash, and doen't want to make Bill any richer. I am platform agnostic. I either have run, do run, or will run, it all. My shop used to run Oracle on Sun. We migrated to x86 and Linux. 9i is a stable product. 11.5.9 is a POS on Linux and Apache. Given lessons learned, next time I'll do Oracle on Windows. Interestingly enough, the latest push from Oracle is to run on Windows. Go figure. I hear all the Linux fanbois going in to techo-drama mode. Why? Linux is more secure, stable? Yeah right... Just look at all of the RED HAT Network alerts that flood your mailbox. Cheaper, Maybe upfront, but over time, cost are about same (Linux fanbois think they should be paid more than their Windows counterparts.) Faster? Hah. Lets talk about real world use where any increase in transaction processing on the back end is eaten by other latencies and bottlenecks in the ether. More stable? Hah. I'got NT boxes that have been running for almost 7 years with no issues. If you know what you doing, the platfom wars are moot. "Slashdot. It's like joining a pissing contest in a wind tunnel"

bollocks

[Stu Charlton]

10g is big, but it's not hard to install, unless you're X11 challenged...