Slashdot Mirror
Widenius Warns Against MySQL Falling Into Oracle's Hands
jamie sends in a blog post from MySQL co-founder Monty Widenius calling for help to "save MySQL from Oracle's clutches." While the US DoJ approved Oracle's purchase of Sun back in August, the European Commission has been less forthcoming. Widenius points out that Oracle has been using their customers to put pressure on the EC, and he questions Oracle's commitment to MySQL, saying their vague promises aren't good enough. He writes:
"Oracle has NOT promised (as far as I know and certainly not in a legally binding manner): To keep (all of) MySQL under an open source license; Not to add closed source parts, modules or required tools; To not raise MySQL license or MySQL support prices; To release new MySQL versions in a regular and timely manner; To continue with dual licensing and always provide affordable commercial licenses to MySQL to those who needs them (to storage vendors and application vendors) or provide MySQL under a more permissive license; To develop MySQL as an Open Source project; To actively work with the community; Apply submitted patches in a timely manner; To not discriminate patches that make MySQL compete more with Oracle's other products; To ensure that MySQL is improved also in manners that make it compete even more with Oracle's main offering."
And why would they.
That's one of the reasons we have open source licenses. So we can fork if we have to.
Best Slashdot Co
Besides being a hippocrite, after he was paid, bolted for the door the first opportunity he got. If it was so important to him, he wouldn't have sold to Sun in the first place. Man up and stay with the company and product if you are so concerned.
So, now, being a very rich guy (1B is a lot of money), he wants to it back for free? That's fair... Right...
It's strangely appropriate that Neo, when he went to see the Oracle to find out that he is The One, was also shown that the reality he was constantly presented with was simply a computer manipulation. This is why "there is no spoon" was such a critical piece of the Matrix puzzle. There may be no spoon, but there can still be a fork.
The Oracle told Neo that he wasn't The One, but the Oracle was lying and just telling him what he needed to hear. The One knows that there is a fork, even if the Oracle leads him astray.
Then there was a whole lot of crap about rogue agents in the system, but the whole movie was clearly an allegory about databases and the GPL.
Has more features, performs better, and has a more permissive license which allows embedding in non-GPL applications.
It's missing some stuff (like multiple indexing types) but I'm sure with a larger user base we could get those features done.
Fork it and rename it. These guys are more interested in "their" brand than the actual code.
Perhaps you shouldn't have sold out to Sun in the first place. Fork the damn thing and let's get on with business.
If he thinks he should be able to call the shots........maybe he shouldn't have sold it.
Widenius tell us sooner?!?
If he gave a shit about what happened to MySQL, he would not have sold it.
Instead, he made gobs of money and no longer has a say in what happens to the property except insofar as he is free to fork it.
Perhaps Monty SHOULDN'T HAVE SOLD the damn thing in the first place if he's so worried about these things happening, no? Besides, there is NOTHING in the world preventing him from forking it, naming it something else and continuing development. NOTHING.
People in the Open Source community have been warning against this for years with MySQL. It is one of the key tenets in the PostgreSQL vs MySQL playbook. Use PostgreSQL because no single company controls the source. It can't be bought. MySQL dug its own destiny by tying its hand into the GPL AND (note the AND) being owned by a single entity.
Get your PostgreSQL here: http://www.commandprompt.com/
How about calling the fork 'MySQL2: The Sequel'?
No one should take his opinion seriously because if he really cared then it wouldn't have sold it. Just fork the thing and forget Oracle.
Maybe he's hoping it would stay open source so he could pinch Oracle's improved code an basically have his mysql money and access to the myql code as it improves so he can plug it into his branch.
As others in this discussion have pointed out, if the concern about Oracle close-sourcing components of MySQL, then why not fork it now?
Also, beyond the large installed user base, is there anything particularly important about MySQL as a database that other open source databases cannot do?
But for me, the biggest frustration is that while there is all this concern about MySQL, the lack of direction is really damaging Sun who make excellent servers (SPARC and x64), software (Solaris 10/Open Solaris with ZFS, Dtrace, Containers etc. etc, OpenOffice, Glassfish, Virtualbox, Sun Cluster (free), QFS/SAMFS (cluster FS)) and many more interesting technologies).
IMHO, the existence of Sun is a positive thing for the open source community and MySQL is a small and largely unimportant part of Sun's inventory.
HisSQL.
> Perhaps Monty SHOULDN'T HAVE SOLD the damn thing in the first place if he's so worried about these things happening, no?
Maybe. I did my quote of shooting in the foot... so me might, too. But now is too late, and if he has awaken now, maybe understanding his creation was in jeopardy... what can he do? Solve the problem alone? That's not how free software works. He needs people -- and for that, he needs credibility. Having sold once, who wants to take chances a second time?
> Besides, there is NOTHING in the world preventing him from forking it, naming it something else and continuing development. NOTHING.
Ah, Oracle is such a fool, ain't it? If there's not a secret agreement, things like branding and communities are not created overnight.
BTW, IMO this is somewhat akin to what is happening to Gnome, but I guess M$ wants to spend less money and do a slower process -- so as to get everyone used to Gnome as an M$ technology.
It's not being sold, it's being assimilated...
I agree that Oracle's dominance and proprietary nature places it in a unique position to dictate terms to its customers. The problem is that Oracle is at least twenty years ahead of all of their competitors in database technology. Oracle 7, ca 1991, has a better overall implementation than the latest and greatest from IBM, Microsoft, PostgreSQL, MySQL, and so on. I mean MySQL is barely out of the 'toy' stage (special purpose applications excluded). In the intervening two decades Oracle has widened the gap. That means for a certain classes of OLTP applications, people tend to think you are suicidal if you recommend anything else.
The only way to minimize this problem is to bring (open source) databases closer to parity, even with where Oracle was twenty years ago. PostgreSQL is the only one that comes close in the open source world. MySQL started out with so many bizarre design decisions and gratuitous incompatibilities, that I wonder if it will *ever* come close, at least not without losing backward compatibility in a big way.
In the U.S. and in Europe, there are regulatory bodies that need to be aware of this potentially serious problem. MySQL is a component of a huge and significant portion of the internet web sites today. What Oracle decides to do with MySQL could have huge and sweeping affect across the entire web economy.
In the interests of preventing any potential large-scale destabilization, MySQL should be forced to spin off into an independent entity prior to the acquisition of Sun. Not only are there competitive interests at play, but a significant component of the internet as well.
Seriously, these MySQL founders have been whining ever since they sold out to Sun.
Please stop. If you're worried about MySQL why did you sell the rights in the first place?
"If a show of teeth is not enough, bite
Agreed. As a big ferinstance, MySQL just barely got two-way replication w/ 5.1, and even then you had to do some seriously weird hoodoo on it to make that happen (hint: it's not a listed feature)... this is a basic function of any full-on enterprise-level DB.
Now Postgres comes fairly close, but everyone else can't even touch it.
If Postgres ever got something resembling the ease and power of RAC, then Oracle would have something to worry about. Until then, they're in a position to dictate whatever terms they want to. (I would've put MS SQL Server as a contender, but clustering that into something resembling RAC is a friggin' nightmare to build and maintain, and I doubt that too many MCDBAs have quite wrapped their heads around using SQL Server on a Core (read: non-UI) install of Windows Server just yet.)
Quo usque tandem abutere, Nimbus, patientia nostra?
besides the obvous potential to fork mysql, there are other options out there such as postgresql.
Much like the line in the song Maria by Blondie, Widenius is blowing hard about something he sold for lots of cash - and the same time he himself has forked MySQL into MariaDB (see what I did there? ;) so he has answered the problem in the way many of you have already suggested. :) ...) ...
No point being down on Oracle btw, they may be a big freaking software company and on that basis alone deserve to be hated, but they did buy a product I am cursed to work with and they have made such improvements in access to knowledge and the software itself I cannot help think that they are not all bad (OK - I know, they ARE bad, but one can dream
What I don't think is that anyone will seriously form MySQL because most people cannot be bothered when there's Postgres (which we all want to be the winner in the mythic and yet totally uneomotional struggle between MySQL and Postgres - wierd, because the struggle between Ruby and Python is so much more emotional but so much less important
Honestly though, as soon as we all dump RDBMS and go for lovely object based databases like Durus the happier I will be
Mod grandparent up. Parent is pretty much a perfect example of the propaganda that Oracle and its (pre)sales minions have attacked insecure and clueless C[IT]Os with for a generation.
This Oracle Tax, much like the Microsoft Tax, hurts the average consumer who has to pay inflated rates for financial services, insurance, etc. when these companies pass on the exorbitant costs of Oracle implementation and maintenance.
Having maintained large infrastructures of MySQL servers for real companies that make real dollars, I'm amazed that there are still CTOs who are insecure about telling the Oracle leeches on sales calls that they have opted to not pay the Oracle tax, and retain some of that money to invest in their business.
I sure wouldn't... and it's a waste of effort to be worried about what somebody else is going to do unless one is actually in a position to influence their decisions.
File under 'M' for 'Manic ranting'
But i take anything Monty says these days with a heavy dosage of salt, especially if it happens to coincide with Microsofts current viewpoint.
http://www.codeplex.org/board-of-directors.aspx
Recommended reading is this from his blog:
http://monty-says.blogspot.com/2009/09/codeplex-foundation-why-is-microsoft.html
Doing a character assasination with those gems in mind would be redundant. Its obvious Microsoft is scared shit that Oracle will undercut it in the SMB market with MySQL and Oracles wast support structure. Call in the drones.
HTTP/1.1 400
I encourage anyone who mistakes Monty for a friend of Open Source to do a little reading...
The case against the case against Oracle-MySQL
MySQL and a tale of two biases
Monty Program AB's Suggestion to EU Commission to Get Rid of the GPL on MySQL
How Many Times Can Monty Sell MySQL?
If MySQL does what you need, at a pain level you can afford, then more power to you. It is certainly a lot more cost effective in those cases. I didn't say anything about ruling out MySQL for any and all possible applications.
MySQL is the Visual Basic of databases - clumsy and of poor quality, used most strongly by people who don't know any better. I would be delighted to see MySQL fail as a project and have its mindshare go to projects that are superior, like PostgreSQL. There are only two things I can think of that the world would miss - MySQLe (the embedded version, which competes with BDB-esque type uses - it's a really cool idea) and the solid Windows support (PostgreSQL added this about a year ago - I'm not sure how solid it is yet).
MySQL's wins tend to be based on good marketing for a bad product.
For every problem, there is at least one solution that is simple, neat, and wrong.
I had moved some new work to postgres, after the sun+oracle announcement I migrated some older code to postgres.
I also weaned myself off of Netbeans.
I've been in this business too long to have any illusions that Oracle is going to fund competitors to its revenue earning products. Oracle
doesn't want you on MySQL- at the minimum they want you on Oracle XE.
#40. I will be neither chivalrous nor sporting. If I have an unstoppable superweapon, I will use it as early and as often as possible instead of keeping it in reserve.
I don't mean to troll, I'm really hoping that someone can explain what the point of all this worry is. Can Oracle really make MySQL not be free? It's my understanding that they could start charing people to buy or license MySQL from them, but that they can't stop people from taking the MySQL source code that's already available, and using it for whatever they want, including selling it. No matter how Oracle changes the licensing terms, or how many proprietary extensions Oracle adds, anyone should be able to take one of the old, freely released versions of MySQL, and sell it for money without Oracle being able to stop them. Even if MySQL is a trademarked name, the new seller would just have to use a different name. I really don't understand what Widenius is talking about.
Monty from Monty Python?
The MySQL brand name and the installed (commercial) user base are pretty much the only reasons for Oracle to acquire the MySQL database business at all. Both worth paying a considerable amount of money for.
No one needs their database server to be called "MySQL" if they don't like the path that Oracle takes with it though. And it would be easy enough to write new (and perhaps better) documentation for a fork.
As long as there was a company behind it, there was always potential that it could be bought. Switch to PostgreSQL. Nobody owns it.
Every implementation involves pain. It's all a matter of how much it will cost you in dollars or sweat to ease that pain, and Oracle makes you pay dearly.
Statements like "Oracle is at least twenty years ahead of all of their competitors in database technology", "toy database" are meaningless and unquantifiable. You may be able to scare the CIO types with that kind of language, but /. is not a fan of FUD tactics
If he cared in the first place why did he sell it to Sun, he has no say he sold it. That's like me selling a car to someone and telling them they can't crash it. If I didn't want it crashed I shouldn't have sold it.
Unless the EC is just hell-bent on obstructing US commerce then Oracle and Sun should just make an agreement to fork MySQL off into a foundation.
What is reasonable and affordable for a small Swedish company certainly isn't for a big behemoth like Oracle. There are now many, many more layers to feed with the product. The time for this decision has passed.
-- Sig down
The problem is that Oracle is at least twenty years ahead of all of their competitors in database technology. Oracle 7, ca 1991, has a better overall implementation than the latest and greatest from IBM, Microsoft, PostgreSQL, MySQL, and so on.
That's a hell of a strong choice of words, particularly when you're on a site full of raging zealots like /. Could you explain exactly what features Oracle has that place it a full 20 years ahead of its nearest competitor?
A.k.a. Ese Culo (that assh*le) in Spanish. :)
This is all about the EU blocking Oracle's acquisition of Sun. They are trolling for testimonials about how the Sun acquisition would force people to buy Oracle DB, which is almost certainly would not:
http://www.moneycontrol.com/news/ibu_index.php?storyid=832
Look at Berkeley DB (on which OpenLDAP uttely depends.) It's now "Oracle Berkeley DB". I don't see any monkey business with that arrangement (although the OpenLDAP people are probably working on ditching BDB just as due diligence.)
Give a man a fish and you have fed him for today. Teach a man to fish, and he'll say "WHERE'S MY FISH, YOU IDIOT?"
whether he's in the right or not, monty is the wrong spokesperson. he has obvious conflicts of interests and possible sentimental ties to mysql. that, and the hypocritical nature of his thoughts as so many others have pointed out.
he'd be better of keeping quiet. he's probably doing more harm than good for his cause.
/. isn't full of 'raging zealots', I see that claimed quite often but I've seen no evidence to back it up, it's a myth ... there are a few, but most people here are fairly reasonable, certainly no more 'zealot-y' than any other online forum, and almost certainly less so. GP is probably a paid shill or works for Oracle. I do suspect there's a fair share of shills here, but they aren't zealots.
Just to make his life easier?
Welcome to the world of commercial open source...
I didn't say "features" (although there are those) but rather "better overall implementation".
For example, with Oracle you can add columns, drop columns, and modify columns while there are ongoing transactions against the table. Try that with DB2 sometime.
MySQL is worse:
That is a trivial example. Generally speaking, however, Oracle gets significant new features with a high quality implementation about a decade before anyone else does. For example, during the 1990s the lack of MVCC and row level locking were serious problems with virtually every database except Oracle. Without them, you can't reliably run large or long running transactions without risking locking every other user out of the database, even if the transactions don't have any row level overlap.
No single entity controls the source of mysql either. It's GPL. If you want to fork it, fork it. You guys are missing the point.
The point is Widenius wants to start a new company, and wants to work off of what mysql, the company (and thousands of volunteers who have contributed to the project) have created over the past N years. He does not care if it goes to Oracle, Microsoft, some made-up nonprofit-ish foundation, or dies. He could really care less about that. He wants to build a company that will make a proprietary product and will make him money.
The thorn in his side, however, is the fact that he can't take the code that was once released as GPL and use it in his proprietary software. He either has to open up his software (which he does not want to do), or else not be able to benefit from all those years worth of effort by mysql AB and others who have contributed to the project.
If the license was just about anything but GPL (apache, BSD, whatever), he could do just that. But he can't.
What, you really think it's all about evil Oracle taking over mysql, and it's not really the license that's a thorn in Wideniuses side? Read a more in-depth analysis by someone who understands the issue a _whole lot better_ than I or just about any of you folks do. Here: http://www.groklaw.net/article.php?story=20091208104422384
Jobs? Which jobs?
before selling your company to Sun. You thought Sun would stick around forever?
Why should Oracle have to offer a product that "competes with its main offering"?
MySQL is a shit database. If oracle tries to crush it, they'll end up with shit splattered everywhere and a stinky hand. It's 100% forkable, and 95% pointless. Postgres, SQLite, and "no-sql" datastores do a better job for most scenarios. And if you're running a PHP-4 script that needs MySQL 4, it doesn't matter who owns mysql.
Do you even lift?
These aren't the 'roids you're looking for.
In the comments on his own damn blog (linked in the TFA), Monty let it slip that he isn't worried about Oracle making MySQL closed source (they CAN'T, or well they can, but anyone can fork and make his own version right now). He is worried about the lucrative business agreements between MySQL AB and business customers willing to pay for a customized version if MySQL for their business use. In other words, he wants a piece of that pie (which he would never get should MySQL AB end up in Oracle's hands). What a greedy prick.
Earth to Widenius come in...
Oracle spent ONE BILLION DOLLARS on MySQL. It has the right to do whatever the hell it wants with it including driving the entire project into the grond.
If you don't like it FORK it. Don't expect any sympathy for being a sellout.
> /. isn't full of 'raging zealots'
> GP is probably a paid shill or works for Oracle
mmmyeah.
He names the db after his daughter, Maria. It's an ugly name for a database.
I don't work for Oracle, and I would like to see other databases to get into the same league inf every material respect. Take Oracle RAC (formerly Oracle Parallel Server) for example, Oracle's shared everything database clustering technology. There are no open source equivalents. MS SQL, PostgreSQL, and MySQL don't have anything like it. Apparently IBM DB2 does, but only in the mainframe editions.
There are often things that can be done to work around these limitations (replication works in some cases, for example), it is just a question of cost effectiveness. There is no reason to buy Oracle just because it is "Oracle". Only if it does what you need better than the alternatives. For many businesses that is the case. Oracle doesn't dominate the business because of FUD. It dominates due to true technical superiority. A business would be positively stupid to pay a large premium for a database that doesn't have any real superiority to much less expensive (if not free) alternatives. That is one of the reasons why it would be great if the alternatives caught up. Transparent clustering for PostgreSQL would be outstanding.
I *can* use PostgreSQL to do everything I could with Oracle 7 back in the early 90s. That is saying something (MySQL doesn't come close). A lot of people don't need much of what Oracle has added since then. If that is the case, there is a great case to be made for using something else. It is certainly a lot less expensive.
I got MySQL 4 to do two-way replication in about 20 minutes, but nice try. Also, speed tests done around 2000 showed MySQL and Oracle neck and neck for large server loads.
Wow, how badly did you screw the pootch in the planing stages if you are changing the columns of a live DB?
The historical name of the language was SEQUEL and was developed at IBM four years before there even was a Microsoft! Your average IBM Yankee, in the EB White sense, graybeard is actually most likely to still pronounce the language name "see-quall" or "sec-well" even though the name was changed by IBM to avoid a trademark dispute.
Oracle certainly has developed a lot of very sophisticated functionality over the years as the functionality of a single standalone RDBMS has become commoditized.
The problem is that a great number of customers that Oracle strongarms into buying their product really don't need what they are being sold, but the thought of using a "toy" database to accomplish your business needs is a hard thing for many CIOs to stomach, especially if it requires you to actually think hard about what you need and what you don't need, and what you can design around.
Take your average medium-sized company looking to build a data warehouse. A pgsql solution using some decent hardware and an entry-level storage array may be enough. For bigger volumes, you can move to a hadoop/hive approach or use something like GridSQL.
For the CIO, unfortunately, there is the axiom "no one gets fired for buying Oracle", and yet-another company is relieved of a good portion of their IT budget.
Non sequitur. He's saying that you *could* do it, and in the enterprise options like this are vital.
From your previous post:
Cite, so I can rip their methodology to pieces. Oracle is for the big boys, son, not you down in the basement. Anything under 32 cores, 256GB of RAM, and a SAN is a fucking joke in the enterprise. And no, Linux/MySQL cannot perform at that level in a production environment. I'm not talking about your sorry fanboi ass scrabbling something together on a one-off university cluster.
Me, I have no skin in this game. Have enjoyed using MySQL on a few home projects running my own web server, but my job is an all-Oracle shop. I just like to see Open Source succeed on general principles.
But even a neutral observer can offer an obvious riposte: Mr. Widenius is clearly not a helpless bug about to be stepped on. Wikipedia says his capital gains hit 16M Euros in 2008, around $25M US. He has that wealth, and another wealth you can't buy with money: He's the famous and beloved Monty Widenius, and open source contributors will listen to him and rally about him in the right cause - he IS a "brand".
Younger programmers, at least, eager for the resume decoration, would probably work for him pretty cheerfully for about $50K per year. A single million bucks a year would get him a respectable programming shop of 20 people rolling. I suspect even with a large project like MySQL, that's enough to do a lot of maintenance and even some development. Then there are those free contributors, which can include whole companies, not just kids-in-parents-basements. If his Oracle-as-quicksand fears are shared, many companies not wanting to end up in quicksand will switch to "NewSQL", or "FreeSQL" or whatever he re-brands it as. Switch, not just their usage, but their code contributions.
I wouldn't allocate a dime to promotion and sales and PR - the best advertising in a move like this is word-of-mouth among server room admins and DBAs.
Needless to say, any SERIOUS challenge, like that, could quickly bring Oracle back to a bargaining table to perhaps concede all those demands he has. They might have more to gain by letting MySQL molder away than by keeping it healthy, but they don't have so much to gain that it's worth the risk of simply losing the whole purchase value if everybody starts bailing to the newer fork.
So he can probably solve this problem for a single million, just 4% of his 2008 capital gains. He's just trying to do it with a blog post, that being cheaper still.
Probably not, no.
There's all kinds of people posting to Slashdot (raging zealots, nonraging zealots, raging nonzealots, and even some perfectly reasonable, smart people), but a good rule of thumb for any Internet forum is: if you are, or ever have, accused another poster of being a paid shill, and actually think this is a remotely plausible explanation for people saying things you disagree with, then you are almost certainly a raging zealot.
Hmm interesting ...
Query 1 = SELECT somecolumn FROM table
Query 2 = DROP somecolumn FROM table
Query 2 returns first with "ok"
Query 1 returns with "fucked if I know, column ain't there no more" ?
How is this useful ?
Story translation: MySQL will be dead soon, like the Pascal programming language.
Add to that that, even if Oracle's pricing model is ridiculous, they never lie about what you're going to get (and you can always get a rebate here and there), and that documentation (even down to their data formats) is ubiquitous and omni-present. Open source newbs tend to yak on Oracle (as they do on postgres) because they won't accept that managing data in large quantities is not something you can learn in an afternoon. Which, in turn, is actually an argument for managing data using btrees and hashes, but that's quite another discussion.
Religion is what happens when nature strikes and groupthink goes wrong.
Online upgrades. Suppose you have a service that needs to be available on a 24 x 7 basis. Is there any reason to shut everything down just because the upgrade script needs to add a new column, drop an old one, or increase the precision or maximum length of an existing one?
We do software as a service, for example, and generally speaking, we don't take our site down *ever*, certainly not for application software updates. Logged in users stay logged in and continue their work without noticing.
This is so wrong I don't even know where to start.
Monty is not beloved - anyone who has been paying attention should be aware that lots of people have lots of problems with him. He isn't an open source hero, but a rather self-serving whiner who has had to have it both ways for a long time.
With the Oracle thing, he's now wrapping himself in the flag of community to argue against the GPL in order to make it easier for him to attempt to take back mindshare on the very product he "sold" to Sun to become wealthy.
He's saying some profoundly laughable things about Eblen Moglen and the GPL in general to do this.
Additionally, paying 20 coders $50k/year != spending $1M/year. Not by a long, long shot.
Also, also, I hardly think doing so would constitute a challenge that "could quickly bring Oracle back to a bargaining table to perhaps concede all those demands he has". I mean, what? That's profoundly silly thinking that, even putting aside the particular ways that Oracle the company operates, is very unrealistic when looking at what the hypothetical Monty the DB Slayer firm could realistically do in, say, a year, and also ignoring what a company the size of Oracle can do in a year. Think about it.
It also assumes that folks in the community will ever want to work with the jerk again. He's burning a lot of bridges, and it may well be that your hypothetical cheap young programmers with know knowledge of the past will be the only folks who would want to work with him. Problem is, green coders are not exactly the type of folk you want thinking about database problems, which tend towards the seriously demanding on all of the reliability, performance and complexity axes.
I forget what 8 was for.
Just curious, since I'm not a DB guy and rarely get to work on servers, why exactly would you WANT to run a non UI version of Winserver? I mean I could understand it way back when where literally every byte of memory counted and sparing a single cycle on a UI could be a bad thing, but today we have so much power, and in the server space even the low end boxes are frankly super powerful, that I just don't really see much of a point.
I mean sure, I can see stripping out WMP and those other pieces that frankly never should have been in Winserver in the first place, but why kill the UI? You can run Powershell with or without a UI, so maybe I'm just not seeing an angle here or something, maybe you can enlighten me.
ACs don't waste your time replying, your posts are never seen by me.
First came QUEL. The followup developed at IBM was jokingly called SEQUEL. It was changed to SEQL and then SQL for trademark reasons. See Wikipedia.
So it was originally called "sequel". Pronouncing it as S-Q-L came later.
GCHQ Quantum Insert installed. If only our tongues were made of glass, how much more careful we would be when we speak
...ForkSQL, the new MySQL.
Ceterum censeo Microsoft esse delendam.
Do you mean master/slave, or two-way replication? There _is_ a difference.
Did Oracle rape and murder a young girl in 1990? I'm not saying it did, but it's been curiously silent on the matter. And if it doesn't plan to do so in the future, why hasn't it promised not to?
Breakfast served all day!
No one is suggesting dropping a column that is still in use. The software upgrade that makes the old column obsolete must usually complete first.
All transactions dependent on the old column must also complete (the wait for this is automatic). Then, either all sessions or all software state that depend on the previous column must go away, OR some form of (updatable) views be used to allow them to continue without early termination (using a column alias or derived column).
Once that happens, it is very convenient for access to the relevant table to continue while the old column is dropped and all its data is reclaimed. Oracle and PostgreSQL can both do this, although reclaiming the previously used space is a second step in the latter.
Until MySQL crashes and obliterates your indexes, yes that's true.
For a site about things like basic rights, Slashdot users sure do like to censor "dissent".
And in addition to all his whining that people here are jumping on, he flat-out lied about his activities with the EC. More discussion on groklaw, the gist is that he has asked the EU to force MySQL to be released without the GPL, then denied having done so. Releasing MySQL back to the community is not enough; he wants a version that he can take proprietary in the future, and that is why a fork is not sufficient.
Oh look! PostgreSQL 8.5.
Once that happens, it is very convenient for access to the relevant table to continue while the old column is dropped and all its data is reclaimed. Oracle and PostgreSQL can both do this, although reclaiming the previously used space is a second step in the latter.
Okay, so far you have one feature which has a better implementation in Oracle than in any F/OSS database (though by your own admission Postgres has already solved half of that problem, and since 8.3 autovacuum has been enabled by default, which at least partially takes care of the second half of the problem.)
That is a great comparison, and contrary to some of the responses, being able to do alter table statements on an in use production system is vital to any serious database solution. It doesn't say anything about Oracle vs. Postgres though as Postgres has been able to do this for a very long time.
I'm not just trying to be contrary here, I would really like to know. What does Oracle have that puts it (20 years?) ahead of Postgres (other than RAC, there were very informative posts above about that).
You seem to have a lot of other people's money to blow away, "son".
I know it must give you a satisfying feeling in your nether regions to see such "enterprise" hardware, but people working for forward-looking enterprises have long-ago figured out that distributed clusters of commodity hardware give a substantially better processing power per unit cost and energy. You just need to turn your brains "on" and turn the Oracle sales calls "off"
Maybe you've heard of some of these "in-the-basement" enterprises... Google? Amazon?
...VB was, before Java, extraordinarily successful. There's no reason to believe that if MySQL went away, mindshare would go to more powerful or more capable solutions, when ease and speed were what sold people on MySQL. People would more likely end up with SQLite then with PostgreSQL.
You're entirely correct. However, Monty blogged a while ago about how that's not enough, and (to reduce things terribly) mindshare follows the original product. He basically said that forks don't survive. He might well be right.
However, he's the guy who sold MySQL. He's the guy that got roughly a BILLION dollars for it, and hence, he's the guy who has no right to whine about the current state of affairs.
"People who do stupid things with hazardous materials often die." -- Jim Davidson on alt.folklore.urban
Online upgrades. Suppose you have a service that needs to be available on a 24 x 7 basis. Is there any reason to shut everything down just because the upgrade script needs to add a new column, drop an old one, or increase the precision or maximum length of an existing one?
We do software as a service, for example, and generally speaking, we don't take our site down *ever*, certainly not for application software updates. Logged in users stay logged in and continue their work without noticing.
The same is true for any serious MySQL site. Just with MySQL, you have to go to the hassle of take out slave, apply change, let slave catch up, repeat for all other slaves, promote some slave to master, apply change to old master to get everything working. This works for changing columns' type/adding columns/removing columns/etc. because MySQL normally uses statement-based replication, not row-based. It can be fiddly, but it works fine. Wikipedia has no downtime for database changes, or any planned downtime at all, and it's a pure MySQL shop. Of course, being able to change tables without this whole procedure is probably quite convenient.
MediaWiki developer, Total War Center sysadmin
Wow, how badly did you screw the pootch in the planing stages if you are changing the columns of a live DB?
Requirements change over time, new features need schema changes. Unless you somehow know exactly what your needs will be for all eternity, you're going to have to change the DB schema sometimes to support a new feature or fix a bug.
MediaWiki developer, Total War Center sysadmin
... *sigh* did they add a clustering feature yet?
But if forks don't survive, why is my Ubuntu laptop running xorg for the display and not xfree86? I think Monty is saying forks don't survive because he simply wishes they couldn't. I'd bet that if the MySQL community wanted to they could fork it and make it stick just fine. Heck, I think that's what IS happening right now, and Monty is putting his fingers in his ears going "la la la la la la".
--- It is not the things we do which we regret the most, but the things which we don't do.
"I doubt that too many MCDBAs have quite wrapped their heads around using SQL Server on a Core (read: non-UI) install of Windows Server just yet.)"
Installing SQL Server on Windows Core isn't supported, though I wish it were.
http://technet.microsoft.com/en-us/library/ms143506.aspx
I haven't tried it, is this something that you've been able to successfully accomplish?
If you post as Anonymous Coward, don't expect a reply.
Notice that there is a DUAL license of MySQL. Whatever EVERYBODY have contributed to MySQL are not squarely GPL but are also the full property of whoever owns the Commercial license.
This mean that Oracle after buying MySQL can take EVERY technology incorporated into MySQL and mix it with Oracle, then leave the project either dead, or simply maintaining it is as bait. If you contribute to MySQL from that point on, you know you are giving away whatever you do to Oracle... In other words. MySQL is dead as soon as Oracle owns the MySQL. GPL only keeps the leftovers.
This barely even rates as a problem. Folks, let's just take out a little simple insurance and remove and operational benefit to messing with MySQL. The MySQL community has the right to copy the MySQL public open source, to a new project branch called OurSQL. This branch continues to receive updates from new versions of MySQL as long as MySQL remains public, and open source. The day Oracle chooses to do anything with MySQL that renders it Closed Source, Crippled, or in any way a Business/Political Pawn to be used against Oracle's competitors, everyone jumps to the OurSQL project and en masse thumb our collective noses at Oracle.
Let Oracle know there is absolutely no value to messing with MySQL, and if they try, will only end up the proud owners of a dead project branch. This protects the autonomy of MySQL, and allows Oracle to intelligently steer the future of MySQL. It also allows those who require that MySQL have a meaningful future as a robust open source product, the certainty that nobody can screw with their future (or at least that it would be most difficult to screw with it.)
I have all the room in the world for Oracle to be whatever kind of for profit, capitalistic, corporation they want to be. I just figure if they might even be slightly inclined to mess with my freedom, or preference, and I have some say in the matter now, I'd be an idiot not to cover my behind while the covering was both legal, and reasonably simple.
A lot of people don't need much of what Oracle has added since then. If that is the case, there is a great case to be made for using something else. It is certainly a lot less expensive.
Thats probaly the case for a majority of people, hence the excessive Oracle price is not worth it for most. Most other DB's are more then sufficient.
"GP is probably a paid shill or works for Oracle."
So if they're a paid shill but don't work for Oracle who's paying them?
I was taking a stab at a guess on it, but I think that the next iteration of SQL Server is supposed to do something of that nature (though to be honest I'd hate to see the command syntax if Exchange is any indication).
Quo usque tandem abutere, Nimbus, patientia nostra?
I have to amplify what sibling (Antique Geekmeister) said... there's a massive difference between master/slave and actual no-kidding 2-way replication. We're not talking two separate DB's here, but a perfect two-way interchangeable (down to the transaction logs) clone of one DB on two distinct servers.
Oracle, PG, and SQL Server can do it... MySQL can barely do it in 5.1 (that is, if you have testicles of iron and dig deep into munging with Federated Tables).
Quo usque tandem abutere, Nimbus, patientia nostra?
"Online upgrades. Suppose you have a service that needs to be available on a 24 x 7 basis."
I can imagine and I see how this is a case you can (and should) control by proper design, not DB*1 features. If you *need* it 24x7 then you can't and shouldn't deppend on a SPOF*2. What will happen when you want to migrate to the new and shinny Oracle current+one? If your DB is not a SPOF then there's no problem with stopping part of the system for upgrading, is it?
*1 DB: Data Base
*2 SPOF: Single point of failure.
Just curious, since I'm not a DB guy and rarely get to work on servers, why exactly would you WANT to run a non UI version of Winserver?
Judging by Exchange, the way things are going you may not have a choice... Right now in Exchange 2k7, half the tools you need to get the job done requires PowerShell, whether you like it or not (given the syntax, lack of flexibility, and Microsoft's fondness for overly-long command names, I'm voting "not". I can do it, but it's like getting a root canal sometimes). I suspect that SQL Server may head down the same road once they get a CLI-version of that going.
Other than that, when you start crawling up into getting that last drop of performance, it's nice to ditch a UI that only the sysadmins see on odd occasion.
Quo usque tandem abutere, Nimbus, patientia nostra?
In the world where PostgreSQL also costs $0, I can fathom no reason why anyone would want to use the glorified hashtable that is MySQL.
Their conclusion, Oracle 9i was on top with 629 pages per second, but "Almost nudging Oracle9i for top honors was the dark horse, MySQL, with 608 pages per second." Not too shabby.
BTW, if you actually can critique this, I would really like to hear it because the methodology sounded great: get a team of developers from the company to implement a webpage that meet certain requirements and then test it under load.
sorry this is the link
You are right, I just use non-transaction friendly MySQL tables--here two way master/slave replication is fine. Transaction are irrelevant for what I do (store lots of information for later reading), so I don't appreciate the difference. Also, this is the article I was talking about.
By the time everybody got their piece of Sun's beeellion dollars, Widenius apparently cleared about EUR 16 million - a very tidy sum, but much less than people seem to assume.
(I would have estimated he got even less than this, until I saw that figure a couple of days ago on his Wikipedia page.)
Is that RMS is somehow batting against the GPL on this one (as covered by Groklaw).
DROP tblOracle FROM mysql; UPDATE tblBadIdeas SET Oracle = true; INSERT INTO tblYourVagina (MyPenis) VALUES ('FUN'); query language is tehfunzors!
'We are trying to prove ourselves wrong as quickly as possible, because only in that way can we find progress.' RPF
This is all about the EU blocking Oracle's acquisition of Sun. They are trolling for testimonials about how the Sun acquisition would force people to buy Oracle DB, which is almost certainly would not:
What does the EU have against MySQL anyway?
There are two choices here:
1) Oracle can support MySQL. If they try to close it, it will fork.
2) Oracle can compete against MySQL.
Conclusion: some within the EU are positioning to have Oracle crush MySQL.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
For example, during the 1990s the lack of MVCC and row level locking were serious problems with virtually every database except Oracle.
I had the impression that "MVCC or no" is the kind of a flamewar that is for DBAs what vi vs Emacs is of the rest of us. As a developer, I'm firmly in MVCC camp myself (I find the concept of isolated snapshots with possibility of conflict on commit much more sane and easier to deal with than the numerous not-quite-isolated transaction models in more traditional DB engines; hence why I'm glad MS finally added it in MSSQL2008) - but, so far as I know, there are some rational anti-MVCC arguments as well, and, as always, you trade one thing for another. Or am I wrong, and it has been settled already?
The only reason he would pull such a show would be that they defaulted on his payment.
Some unforeseen circumstance has caused Monty to lose his full monty.
Besides he has moderation turned on his blog.
I'd look for alternatives if you need sql.
I don't care about Montys ethics or money. MySQL has always been dual license. Oracle doesn't want to give anyone anything for free. MySQL is a database that eats away at their bottom end. I have no doubts that they talked about MySQL when they were talking about buying SUN. I can only expect that Oracle will try to do with MySQL what Microsoft did with FoxPro when they bought it from FoxSoft: the very next version of FoxPro suddenly could only access 1% of the data as the previous version. Several years and many releases later, it could nearly access as much data, but in 1/10 the speed, and with significant other bottlenecks. in short, they bought it to kill it. MSSQL won, Foxpro lost and microsoft decided. They got their money back in 1 release (MSSQL was suddenly more expensive). I care about MySQL. I plan on continuing to use it. I (me, Mr. Anonymous User), will so fork it if I have to, dammit!
He already sold it all off to Sun, once, for a pile of cash, why does he care now about which corporate owns it?
I'm sure if Oracle cut him a check that had enough 0's on it, he'd shut up - he's already demonstrated that he values money in his pocket more than open source purity, so I'm sure stuffing his wallet a little bit more would pacify him.
Hyperbole aside, yeah, Oracle is in a league of its own. That's exactly why the idea that Oracle "fears" MySQL and would want to squash it is so utterly ridiculous.
It makes a lot more sense to me that Oracle would acknowledge the fact that not everybody needs or wants Oracle db, so they might very well turn MySQL into a low-end offering, thus catering to a wider audience.
OK, so they might make it close source and commercial only. I highly doubt it because they'd lose the network effect that MySQL currently enjoys, but let's say for the sake of argument they would. So what? They bought the copyright fair and square, it was sold to them. If Widenius didn't want that he shouldn't have sold. And the source is still out there and it can be forked so I really really don't see what his problem is.
On a side note, MySQL is quite lame on technical merits. I suspect Oracle would have been much better off buying something like Postgres. Except that the Postgres copyright is not for sale.
i ate crayons when i was a kid and now i have two braincells and the blue ones taste nicer
RAC is the big one, the one that would take a *lot* of work to make something comparable to. In general though, most things fall into the category of quality and depth of implementation.
To take something simple, numeric (NUMBER) types have a much more efficient storage format in Oracle than PostgreSQL. Much faster the last time I benchmarked them as well. The efficiency is a significant problem for financial applications. That could probably be fixed relatively easily, but it would still be a matter of catching up to where Oracle was twenty years ago.
Another one is type compatibility. PostgreSQL historically has had *big* problems with treating various numeric types and integral types as equivalents of each other. Last time I evaluated it you had to use typecasts all over the place to make sure that indexes were used.
Last I checked PL/SQL was much faster than PL/PGsql. PostgreSQL doesn't automatically create updateable views yet. AFAIK, there is no way to make a view filter on some sort of session variable, so you can create virtual private databases.
Oracle has no end of other relatively exotic stuff that makes a difference to some healthy percentage of their customer base. We don't use five percent of it, but its there and it is the sort of thing that would take a significant amount of time and effort to duplicate, more due to breadth than depth in this case. More power to the folks who are closing the gap.
I use Percona MySQL builds for over a year already. Heartily recommended. And no, I have no connection to Percona whatsoever.
On 60 Minutes tonight, Andy Rooney did the richest 7 Americans. On number 3 he goes, "I have no idea who Lawrence Ellison is."
Here's a tip, Andy. Larry Ellison has no idea who YOU are.
Also, Steve Kroft looks like an alcoholic. Nice try quizzing Obama on his agenda when you have a fifth of Jack hiding under the chair. Douchebag.
But anway, in the top 7 richest Americans, two were software (Microsoft and Oracle), one investor (Buffett), and I guess the other four were Walmart-related. Kind of sad that Buffet couldn't beat out Gates but I suppose neither one is clubbing and wearing the latest fashions anyway.
Oracle has announced a statement today making commitments concerning MySQL that may (or may not) address some of these concerns -- of both Widenius and the EU.
http://www.marketwire.com/press-release/Oracle-Corporation-NASDAQ-ORCL-1090000.html
These include:
* Continued Availability of Storage Engine APIs
* Commitment to enhance MySQL in the future under the GPL
* Support not mandatory
* Increase spending on MySQL research and development
* Continuing to maintain the MySQL Reference Manual
* Preserve Customer Choice for Support
And some other things about preserving the conditions of licenses currently held by storage vendors.
Healthy skepticism is of course always a good idea. On first reading, I can't tell how binding these commitments are (the statement says "Oracle hereby publicly commits to the following", and that's about it), and it doesn't exactly make Widenius' commitment to the timeliness of new releases and patches, except for the commitment to increase spending, which Oracle presumably would like to have result in new revenue.
But Oracle is evidently trying to address the EU's concerns in an effort to get the deal approved, and the EU might get them to make these commitments binding. The EU's initial reaction appears to be positive:
http://www.bloomberg.com/apps/news?pid=20601087&sid=a4SRxTHKHzTA&pos=7
Always keep a sapphire in your mind
I've been using Oracle for the last two and a half years, after a year or two of 50/50 Oracle and SQL Server at another company (with occasional MySQL and Informix). Before that, I did a whole bunch of SQL Server 7 and 2000 work for about six years.
We made schema updates all the time in SQL Server, and never took the database down for it.
You mentioned "long-running transactions" in another post as an Oracle advantage, which I suppose is true, but the SQL Server solution of "Don't do that." really isn't that much trouble. You arrange your T-SQL to get all the answers into temp tables or table variables before starting a transaction to avoid locking trouble, and then discover it makes testing easier, anyway. I'd do it in Oracle, too, if temp tables weren't such a nuisance. It's certainly not enough of a problem to outweigh mutating table errors.
Eben Moglen just released the article The European Commission and Oracle-Sun.
I am not suggesting you have to take down the database, but rather that you either shut down applications, or queries and/or updates block until the table modification (or some other unnecessarily competing transaction) is complete. That is a serious problem in some applications. It is those sort of limitations that practically put Ingres out of business.
In Oracle, if you want strict consistency for a report you do "SET TRANSACTION READ ONLY" and for the rest of the "transaction" you get a read only snapshot of the database. Otherwise you generally get statement level consistency. New queries see newly committed data.
The big advantage of MVCC is that it prevents writers from blocking readers. The big advantage of row level locking is that it prevents writers from unnecessarily blocking other writers. That goes doubly for DDL operations. In Oracle, for example, you can create and rebuild indexes while the indexed columns are being updated. If the index creation would other wise block transactions for several minutes, that is a major advantage.
I'll take "Things I should have thought about BEFORE selling MySQL to SUN" for 1000 please Alex
For example, with Oracle you can add columns, drop columns, and modify columns while there are ongoing transactions against the table. Try that with DB2 sometime.
With PostgreSQL you can run DDL commands inside a transaction. Try that with Oracle sometime. ;)
Another interesting (but proprietary) project is Maglev. The idea is that you write programs in Ruby and the virtual machine gives you persistence and clustering for your variables in a transparent way.