With today’s surprise announcement that Oracle will acquire Sun Microsystems, several questions were raised as to some Sun products, including MySQL, Solaris, and OpenOffice.org. Browsing around the net, there are several viewpoints on the future of these Sun products, and the OpenOffice.org team has even issued a statement itself.
Let’s start with the OpenOffice.org team. They are optimistic about this announcement, and are happy that this marks the end of the uncertainty around Sun’s future. “We’ve been very comfortable with the way that Sun set up the project and has kept a careful eye on us over the years,” John McCreesh, OpenOffice.org’s marketing lead, told ZDNet, “They’ll be a hard act to follow, but we’ll approach the future with optimism and an open mind.”
A more problematic situation could arise for MySQL. What would Oracle, a company focussed on database products, see in the open source database MySQL? Software engineer Ryan Thiessen, an 11-year MySQL veteran, will be speaking at the MySQL Conference this week, and he wrote a blog post this morning, titled “Stunned”.
I am stunned at the news that Oracle is buying Sun. This is not because I fear change or uncertainty, last time this year I was cautiously optimistic about Sun’s purchase of MySQL. But not this year, it’s fear and disappointment over what this means for MySQL.When I read this as a rumour a few weeks ago I thought it was a joke of an idea. Why would a high margin software company want to buy a declining hardware business, even if that hardware is great? As for their software, I cannot imagine that Oracle is interested in Java, MySQL, etc as revenue generating products, it would just be a tiny blip for them.
Another line of thinking goes into another direction than Thiessen’s. Oracle serves the higher end of the database market, compared to MySQL which focusses on the lower end of the same market. With round and about 11 million installations, MySQL offers Oracle a lot of leverage in a market they previously didn’t really took part in. “While no one could ever rightfully claim that MySQL threatens Oracle’s higher-end database offerings, its addition to the portfolio gives Oracle additional leverage in a market with significant growth potential,” Independent analyst Carmi Levy told BetaNews, “The MySQL installed base of approximately 11 million gives Oracle sales teams fertile opportunity to have conversations they haven’t previously had.”
Personally, I fear the future of the SPARC platform. SPARC hasn’t exactly been doing stellar lately, and has all but vanished from the workstation segment. In the high performance computing market, it isn’t doing very well either; in 2002, 101 of the Top 500 supercomputers were SPARC-based – now, that has dropped to just one. Developing and maintaining your own architecture is a costly business, especially in the face of cheap x86 processors.
As a final note, both IBM and Microsoft were surprised by the announcement. “I just learnt it […] I need to think about it. I am very surprised,” Steve Ballmer told Reuters. Apparently, IBM was surprised as well; they were still hoping to buy Sun.
Oracle would lose the most if Sun went away and IBM would probably gain the most in their database business.
I would imagine that MySQL will continue, simply because Oracle is bad at the low end and Java will continue because it has momentum. However, I’d think that they’ll both be spun off to separate subsidiaries.
is it really a surprise?
“I am very surprised” said Steve Ballmer
“According to a source of mine, IBM hadn’t given up on purchasing Sun and was blindsided by Oracle’s move” said Ina Fried
the answer to your question is yes
(the sun buy signals a big change in oracle’s business, going from NOT a vertical all-in-one into a vertical all-in-one. it was always something they could have done, but only internet oracles saw it coming in hindsight)
That’s PR for saving face.
Practically all of Oracle’s enterprise software stack is built on Java. From their server management tools, to AP-GL, to PeopleSoft. I seriously doubt Oracle will spin off Java. They are probably quite delighted to be gaining ownership of it as well as getting Sun’s permanent seat on the JCP, which comes with more control over the future of Java than other companies have.
In the past, Oracle has sometimes had to maintain their own version of Java to work around bugs in the Sun version. They won’t have to do that anymore.
Edited 2009-04-20 21:30 UTC
Well said.
In fact, in terms of Solaris, Java and SPARC – Oracle’s purchace makes perfect sense.
They can now be the Apple* of enterprise databases.
It’s also not contary to their existing business model.
* in terms of controlling the system at every stage.
Not to mention that Open Office also uses Java and will fit in nicely with everything else in terms of offering an end to end software/hardware stack.
Now everyone but Oracle will have to do that instead
LOL
I don’t see it damaging OO in any way. Not too seriously though. Wasn’t the last version of JAVA GPLed?
I see this being good for Oracle because they can be more of a competitor in Enterprise development. Oracle + Java vs MS (SQL) + .NET !
i think that btrfs and oracles linux-endeavor will be the first to fall
Well, Oracle could kill the investment in btrfs, but that doesn’t means the fs will die, it has people from other companies working on it. And the main authors of the proyect can quit oracle and search a job in other linux company – that’s the beauty of opensource.
Fortunately Chris Mason just send an email saying that oracle is going to continue supporting btrfs.
Cannot for the life of me see MySQL remaining open source & cost free. It’s instant revenue for Oracle. They will likely fork the code & enhance with proprietary extensions, then quietly go about up-selling to the 10 million or so existing MySQL users.
OpenOffice? What use has Oracle for an open office suite? Perhaps they will leave it alone as a PR exercise.
The Sparc is dead. Long live the Sparc! Oracle may have a use selling pre-built turnkey systems though…
Good luck close sourcing mysql. Mysql’s founder has already set up a separate business around a fork of mysql. If they completely close source the main branch, where do you think everyone would go? It should be noted, however, that Mysql currently does sell the opportunity for its partners to close source their modifications to mysql. They might put more emphasis that section of the business.
Yeah, I have to agree. I think a lot of companies are starting to back off from their open source commitments. First, it’s not really that profitable for them, and it’s not as much of a PR coup as it used to be. Adding to that, with a recession in progress, there’s not a whole lot of money available for altruism.
I’m thinking that the best outcome you can expect from the Oracle acquisition is that they’ll release the things they can’t sell for a profit into the wild. Other than that, I expect they’ll to let them quietly fade away, the way Apple did with Darwin.
Realistically I think the open source movement is starting to fade, at least as far as getting any corporate love goes. There will still be open source projects, of course, but I think corporate sponsorships are going to be a little harder to come by.
I’d also agree about Sparc going away. It’s been an obviously moribund platform for quite some time, and I could easily see Oracle promoting an Intel Solaris-based appliance for their low end customers, and thus avoid competing with partners like HP and IBM on the high-end RISC boxes. Continuing to develop Sparc just doesn’t pass muster from a cost/benefit perspective.
Edited 2009-04-20 23:31 UTC
There is precious little evidence for this. Open source development continues apace. It has an estimated 1.5 million developers/testers/other people involved worldwide. It comes out with new enhancements and applications and versions at a frenetic pace, compared with the closed source competition. It has companines buying up open source projects all over … huge software companies such as IBM, Novell, Nokia and now Oracle.
Mozilla are a non-profit organisation, but they make pots of money from Google in order to easily fund further development. Mozilla’s firefox browser (by some accounting anyway) is starting to approach 50% of the browser market on its own.
Speaking of Google … they also develop open source code of their own, and they fund the Google Summer of Code.
OpenOffice is estimated to have about 20% of the business office suite market, and rising, and who knows how much of the home dektop market. It has had over 100 million downloads, and who knows how many installs from each download.
IBM owns the mainframe and supercomputing market, and open source is a significant presence in both of those arenas.
Linux owns the embedded market, and any number of companies are invested in that.
Open source movement starting to fade? I don’t think so … rather the exact opposite. It would appear to be gaining momentum all the time, and starting to burst out everywhere.
Microsoft appears to be busily playing whack-a-mole all over the globe desperately trying to stop it, and apparently giving away all its stuff for free in the process.
http://www.itwire.com/content/view/24477/1231/
huh, Dead alive?
Call a doc, Mysql is in the ICU !!
I’m curious/nervous what Oracle will do with Virtualbox. Free, multi-platform virtualization doesn’t seem to fit with any of Oracle product lines. I really hope it doesn’t get lost in the shuffle…
Same here.
The Open Source Edition is licensed (mostly) under GPLv2 so it should be possible to a least fork it if Oracle decide it does not fit in.
Double post, my apologies.
Edited 2009-04-20 21:49 UTC
Virtualization goes with databases or at least it should. Where I work our IT dept just upgraded versions of Oracle on a machine that was hosting a handful of databases that belonged to different orgs. That was fun getting everybody ready to swtich all at the same time.
The thing with databases (especially high performing ones) is that you want them as close to the metal as you can get. Virtualizing databases only make sense if they are infrequently used, or have a small amount of users.
VirtualBox wont be killed, prepare it to get a huge boost instead! It wont be closed sourced for most part. I think Oracle will develop it to be a direct competitor to VMWare. Remember that most of the VMWare products are free to use and you have to pay for support. I really dont see them trying to offer VirtualBox for a fee against VMWare ESXi which is free.
Dang it… I use MySQL for pretty much everything I do.. oh well.. PostgreSQL here I come!!
Although I’m more preoccupied with VirtualBox as I use it quite extensively. It is true that it does not fit in Oracle’s business model.. but does any of the Sun parts fit? I see that they are just trying to offer more products and increase their footprint in the market… that’s my 2 cents..
For many projects you can use SQLite. It’s bundled into PHP for instance. It works great and it’s super fast.
And uses database level locking. Not exactly going to be super fast if your site becomes popular.
I would not build a public facing Web site that uses SQLite as a backend. If your site takes off, you will find yourself having to migrate to a different database.
Fr intranet sites though that are only visited by a known number of people, then it can work great. But for a public Web site? Don’t do it.
Dude, never, ever, ever use SQLite as a backend to a multi user system. It is great as an embedded db for single user apps, or as a test database for unit tests, etc. Anything more then that, and you look elsewhere.
MySQL for small-ish sites, or when you don’t need transactions or triggers (i.e. when you can get away with MyISAM). If you want beefier db features (like transactions and triggers), Postgres will take you a long way. If you have a budget, and find that support and developer friendly extensions are worth the price, you look at the mid tier databases, like Sybase, DB2, or SQL Server. If you need extremely high performance for ridiculously large datasets and cost is not an issue, Oracle is the only choice.
What is ridiculously large dataset, these days? > 1 petabyte? An exobyte?
Also, why wouldn’t you consider MySql-Innodb for loads that need transactions? Everyone else does. That’s probably the most popular combo.
You’d really consider Sybase over Postgres? Interesting. I wouldn’t have thought that many people would consider sybase for anything new.
Because Mysql sucks?
MySQL bashing is quite popular, and there are some valid hits to be made against MySQL. But it hardly “sucks”.
With the way MySQL is often portrayed in comments sections of articles liek this, you’d think it’s slow, riddled with problems, unstable, and utterly incapable of handling even basic DB duties.
Yet hundreds of thousands of dynamic websites run MySQL as the back-end database, and do so very well. WordPress, various popular message boards (even this site?) run MySQL. MySQL was relatively easy to work with (especially if you’re an application developer, and not a DB administrator, which happens quite a bit in the roll-your-own and small-medium sized crowd), the documentation was much better and more accessible than other projects for a while (PostgreSQL caught up not too long ago).
MySQL works for what it’s used for, and it works great.
You are correct, it doesn’t. *I* think it sucks.
The popularity argument is great in that you can use it against things you dont like and for things you do like.
Funny, I always found PostgreSQL’ docs much better than MySql’s.
It’s not a popularity argument (which I agree are stupid, as stupid as pro-obscurity arguments). It’s evidence that MySQL is highly effective at what it does, and that tens of thousands of developers find it fantastic to work with, so much so they base their projects on it.
Well, you’ve got the one manual and a wiki on PostgreSQL: http://www.postgresql.org/docs/, and you’ve got all this at MySQL’s site: http://dev.mysql.com/doc/index.html
Of course it’s a judgment call, but with that and the huge install base, sites, blogs, and so forth dedicated to MySQL configuration, maintenance, and programming, I think one can make a pretty good argument that MySQL has the upper hand in documentation.
Having a lot of it doesnt mean it’s good.
I’m sure you an also make the opposite argument.
Would you mind explaining why you think it sucks? Then maybe we can have an intelligent conversation about database use cases. Instead of a grade school argument about who’s dad can beat up who’s.
I’m too tired for the whole shebang but lets just say that the time it took for mysql to get transactions (and for Monty to drop his bullshit “you can use table locks instead” argument) spoke volumes about how much these people understood RDBM’s design.
Fair enough. Most people do not use Mysql as a traditional RDBM. Many people have found that they don’t need all of the features of an oracle like RDBM when dealing with web run databases. This is one of the biggest misunderstandings when people look at Mysql, and why some of them look at it and call it a “toy” or that it “sucks”. It doesn’t do what they expect from a database. But as it turns out there are cases in which you have use cases that are a bad fit for the traditional RDBM.
Re Monty’s comment, One of the nice things about open source, is that you don’t have to toke the companies official word. You can try it out yourself, and see where it performs well, and where it lacks. There are many great resources on the web for examining how people use mysql to achieve amazing results. I’d love to learn more about similar success stories form the postgres camp, other than “it rocks” or “it doesn’t suck as much as mysql”, for comparison.
Perhaps.
Or they think they don’t. Lets face it, most web developers have no idea how to design a database correctly. The fact that so many don’t even understand why transaction isolation make their life easier and their apps more solid is a testament to this.
True enough. Most of the time developers seem to use MySql as something of a glorified flat file storage.
There might be better solutions for that use case though, such as memcached or Hypertable.
Edited 2009-04-22 04:31 UTC
The great thing about platforms like MySQL and (PHP/Ruby/etc.) is that they work for all levels of developers, from beginners writing their first applications to world class developers who write ridiculously popular apps like WordPress.
I don’t see how anything you’ve brought up so far would give a developer reason to want to convert from MySQL. I’d love to see actual examples of why a developers life would be easier if I gave up on MySQL and moved my development to PostgreSQL or other non-MySQL databases. Instead, all I see when MySQL is brought up are talking points that most of the people making them understand only at a very superficial level (not saying that’s you, but if you’ve got specific examples, I’d be genuinely interesting in hearing about them).
I don’t think denigrating developers for the choice they’ve made or talking point attacks are going to convert anyone.
So does, for example, PostgreSQL. It’s in no way harder to install PostgreSQL on Ubuntu than it is to install mySQL.
I’m not trying to convert anyone.
Sorry, did I say it would be easier? I said transaction isolation makes it’s easier which, after way too long, mySQL actually has these days.
I think the entire argument “It’s not a true RDBM” argument is fairly pedantic. My answer is: I don’t care if it adheres to what someone’s notion of a what a RDBM should be. I don’t care if it’s not really an RDBM. It does what I need it to do, it does it fast, and it does it reliably.
Just my opinion, but the “true RDBM” argument isn’t really going to convince people to not choose MySQL for their project.
It’s like the “layer violation” argument against ZFS. It’s pedantic. So what if it is a layer violation? Does it work? Is it useful?
True RDBM’s or not, having transactions is an important feature, especially for stateless web applications. What happens when the user presses the browser stop button in the middle of your websites database update and you dont have proper transaction isolation? Incomplete and inaccurate database content, that’s what.
Err… No. Pressing the stop button on the client side does not affect the code running on the server. If the user hits the stop button before they finished sending the request, nothing will happen on the server. If the user hits the stop button after they finished sending the request, then the server will process that request in it’s entirely, regardless of whether the client receives the response.
The only way this could result in data corruption is if you’re doing something stupid. Like having a multi-page process where data in the database is updated on each page, rather than doing the updates all in one go.
If you’re doing that, you’re an idiot, and deserve whatever mess you get. Transactions will not help here.
That said, transactions can be very useful in web applications. I certainly won’t deny that. It’s just that many web developers wouldn’t have a clue about how or where to use them, or they’re using a framework that neatly hides the database behind an abstraction that prevents you from being able to use transactions. Or enforced referential integrity.
Hmm…are you sure? I’m not so sure that when the connection is lost the process for that session isn’t abruptly interrupted. I guess it would depend on what application framework you’re using but I’m almost certain that PHP apps will crash and burn if you interrupt them.
And that’s the problem now, isn’t it? Lack of understanding of how to use the tools and how to do good designs with those tools.
(And lets not even start on database designs that use autoincrement integer id’s in every damn table)
Ah yes, another great feature misunderstood, if understood at all, by many, many web developers.
Well, funny thing… Transactions can be useful. There are some situations where you cannot do things reliably without transactions. However, most obvious cases for transactions can be replaced with multitable updates , deletes, or insert on duplicate key update statements that make it faster than using transactions, and reduce the risk of deadlocks.
Sort of but transactions exist so that you do NOT have to do multitable updates and such. If your app crashes (or for a webapp, connection is lost for whatever reason) in the middle of a multitable update you’re a bit screwed. If you have tables with relations there’s no reason in earth you shouldn’t use transactions. The small speed penalty hardly makes a difference in this day and age.
Is it a “layering violation”, or is it a re-thinking/refactoring of the layers? ZFS is still a layered storage system (vdev -> storage manager (pool) -> volume manager -> filesystem). It just uses different layering than the standard Linux storage stack.
But, it’s not like it’s impossible to make it work on a “normal” storage stack. Just look at the FreeBSD port of ZFS, which uses GEOM (the FreeBSD storage system) and didn’t require ripping huge chunks out to shoehorn ZFS in.
There’s an interesting blog post about this here:
http://blogs.sun.com/bonwick/en_US/?cat=ZFS&date=200705
Ignoring the trollish bent to your post, there are some obvious use cases that do not perform well in Mysql right now. I was asking for a more intelligent, informative answer from someone who might have some real world experience in evaluating postgres & mysql for various workloads.
MySQL is great if you just need a place to stick a low volume of un complected data. It is better then a flat file, and if you go MyISAM it is very fast (mostly because it isn’t doing much).
I like using it for cases when I just want to bang out a quick, uncomplecated site very fast. Although since heroku exists, I tend to just dump those sorts of sites there now (and they use postgres).
Well seriously large databases means data warehouses corporate databases, everything where data security is priority #1. That is where data security is more important than money and there Oracle comes into the game offering a ridicously good database for ridicously high prices, and they know why they can charge that much… I would take Oracle for other things as well, but it is a money issue, the database simply is not affordable to be used for smaller tasks!
Why I would not use MySQL Innodb for loads simply is speed and data reliability, especially if you take parallel access from multiple users into the equation, this combination simply is slower than postgresql the more complex the tables and joins become. MySQL only can really shine speedwise in MyISAM but then even a file access can be faster as MYSQL without too much data security loss 🙂
As for Sybase over Postgres, I personally think Postgres is in the same league as Sybase maybe better (Don´t know the state of sybase, but Postgres is already very advanced in its state and features)
I’m not sure about the 5.4 release, by InnoDB has sucked for a very long time. There wasn’t really any non-hype reason to choose MySQL InnoDB over postgres.
I wouldn’t, but people do. I like postgres and mysql just fine, and generally choose one of the two for anything I do myself. I have worked with oracle before, and from a dev point of view I hated it passionately (I know PL/SQL, but it is never going on my resume). I started working at a MS Gold Partner about a year ago now, and SQL Server has been really impressing me from a dev point of view (tooling wise and T-SQL is my favorite sproc language I have used so far).
There’s actually no reason you can’t use it for small intranet sites and such. It is transaction safe, and it’s even thread safe if compiled with thread safety. The problem is simply that it locks the entire database during reads and writes. But for say, five people accessing an internal Web application at a small business, this is not going to be a problem. The database lock durations are going to be short enough that no one is going to notice in most common usage scenarios–especially with good query caching, and if most operations are selects. But with large numbers of users hitting the database often–especially if there are a lot of write / update operations, now the performance could be a problem.
The problem basically comes down to the fact that it won’t scale well if your site grows popular.
However, it should NOT be used in a situation that would involve something like having multiple clients connect to the same database over NFS. This could result in corrupt data because many OSes have broken file locking mechanisms.
Edited 2009-04-20 23:06 UTC
The thing is, you are basically shooting yourself in the foot for no reason. MySQL is free, and while it is harder to set up then SQLite, it isn’t exactly complected.
MySQL or Postgresql at the low end, sure. SQLite for single-user applications, sure.
Not that it is my area, but what is wrong with Firebird at the mid-tier though? It supports transactions and triggers.
http://www.firebirdsql.org/
http://en.wikipedia.org/wiki/Firebird_(database_server)
You don’t need a budget for that.
http://www.firebirdsql.org/index.php?id=about-firebird&nosb=1
Edited 2009-04-20 23:18 UTC
For the first time ever I agree with you, Firebird rocks.
Why the hell would you put PostgreSQL at the low-tier and not the middle-tier?
Seriously, please explain that rationale. I’ve got to read this one.
Edited 2009-04-21 01:54 UTC
Honestly, I’ve never used firebird or come in contact with someone that has, thats the only reason. I wasn’t trying to write a cannonical guide to these sorts of decisions, just point the OP in a better direction.
Sorry, SQLite does not play in the MySQL space just as MySQL doesn’t play in the Oracle space.
I wonder what will happen to the Gnome accessibility efforts? Sun was the largest driving force behind them, and employed a few developers towards those ends.
Not that it affects very many people here, but it’s a concern for me. Fortunately all the work put into it has been open source, so it’s not likely to die, but hopefully someone will pick up and fund the effort, or at least provide it with direction. Canonical would be my guess as to who picks it up, if anyone does. I don’t see Oracle continuing with it, GNOME projects really aren’t relevant to their business.
Oracle operates at the extreme high end, we are talking 50k/cpu high end (for the enterprise version). MySQL is just this side of not even being called an RDBMS, it lives on the extreme low end, and doesn’t really do that much.
Remember, Oracle bought sleepycat (who own and dual license berkleydb), and have continued offering it as open source, while selling licenses to those who don’t want to abide by the open source license. I would imagine the same thing will happen with MySQL
– Oracle is not interested in OO.o. IBM failed with SmartSuite and Lotus Symphony, why should Oracle repeat this mistake?
– Oracle is not interested in MySQL. It’s not a remote threat to Oracle’s database. And it’s no source of income (MySQL AB was sold to Sun for ~1 billion USD, how much revenue could they bring per year?),
– Oracle is most probably not interested in SPARC. I find it a bit sad, but I’m pretty sure SPARC will be killed right after the acquisition is completed,
– Oracle is somewhat interested in Solaris. However even Oracle cannot afford to treat other systems (Windows, RHEL, other UNIX variants) as second class citizens. So all the other OS’s will still get highly optimized versions of the Oracle database and related products,
– Oracle is somewhat interested in ZFS. OTOH the Oracle database is to some degree an operating system and a filesystem by itself. So maybe some of ZFS’s underlying technology (and patents) might find their way to the Oracle database,
– What Oracle is really interested in is Java. After they acquire Sun they will control 3 of the top J2EE application servers on the market (their own, BEA’s and Sun’s). They will have control over where Java is heading. And they will make good use of that control.
And after all – 7.5 billion USD is just about the yearly operating income of Oracle. So even if it goes wrong, it’s not going to be that much of a problem
.
– Oracle is somewhat interested in Solaris. However even Oracle cannot afford to treat other systems (Windows, RHEL, other UNIX variants) as second class citizens. So all the other OS’s will still get highly optimized versions of the Oracle database and related products,
I am not in the least bit knowledgeable of database markets or anything, but atleast my understanding says that Oracle is a lot interested in Solaris. Imagine, a whole OS they can specifically tailor from head to toe to provide the maximum performance and features for their database installations. Slap on all the necessary administation, recovery and other tools and you’ve got a solid package you can sell to customers.
they would’ve done something like that with their oracle unbreakable Linux if they foresaw any advantage…but now that they have a hardware platform of their own it sounds feasible. Now what i would like to see is a merging of the almighty RMAN with ZFS/BRTFS.
Ah, yes. But you miss one important detail. Unbreakable Linux, being essentially Red Hat Enterprise Linux re-branded, is still governed by the GPL. Now that they own Solaris outright, Oracle could potentially offer a closed-source that’s very closely coupled with their database without the risk of having to disclose any of the code. Furthermore, Oracle really couldn’t make any radical changes to Unbreakable Linux without breaking their compatibility with RHEL, which was kinda the whole point in the first place.
I’m not suggesting open source Solaris is going away; simply that Oracle may have some plans for building a much more tightly integrated platform than they ever could with Unbreakable Linux.
That said, I agree with many of the other posters here who suggest that Oracle’s primary motive for the acquisition was greater control over the future of Java.
Oracle has previously made three different offers to purchase MySQL, originally at $300M, then $500M, and most recently (2007) $850M, so I wouldn’t dismiss their interest. They even purchased InnoDB to try and force a sale.
MySQL can give Oracle a footprint that their enterprise-class offerings simply won’t fit into, and by controlling the direction of MySQL, they can ensure a pathway from one to the other as customer requirements scale.
It wouldn’t make sense to close it off, it’s really not in the same class as their core offering so wouldn’t be cannibalizing anything. Rather, by keeping it open they can help ensure that it remains a popular choice for db requirements. Even if they may not generate a revenue stream from more than a small segment of that userbase, they can help ensure that MS and others don’t as well.
Just my speculation.
I don’t think they’ll be able to “control” mysql’s future if it isn’t in the best interests of the general mysql communities. If they try to stunt its growth or limit its scalability on more powerful hardware, the community will step in support a fork that does.
So apparently if something doesn’t work in one year and bring in billions upon billions of dollars whilst simultaneously bankrupting Microsoft it is apparently a failure?
So you equate poor marketing, a lack of investment and terrible leadership and direction by the owner as equating to a poor product? interesting to see you’re willing to make spontaneous leaps on no evidential basis.
Why? Sun is focusing on the multicore monsters (UltraSPARC RK) that are the right price and perform well – and profitable too. Their SPARC machines are being replaced by re-badged Fujitsu SPARC64 VII. Oracle has always wanted to control the whole stack and extract more profit from their customers – price hikes only work for so long.
Who said anything about neglecting the other platforms – but what we will see is Oracle investing heavily in OpenSolaris to fix up the short comings that make the platform unattractive to their customer base.
True, then there is running Oracle on a raw partition as to avoid the overhead of the file system.
Which is great because there will finally be an organisation willing to take Java by the horns and map a direction rather than whimsically swaying according to internal politics. Swing versus SWT is a prime example of Sun politicking pushing something they know is inferior but do so because they (Sun) created it.
I don’t see any problems. 2 years it’ll be fully digested and you’ll see a ‘database in a box’ being sold. Roll it it, hook it up to the network and power supply, flick the switch and you’re ready to rock. Companies want hardware and software to hit the ground running and most can’t afford having this hardware sitting on the side lines as the tech boffins are trying to get it working with the existing setup. Oracle will supply that solution – you might even see more services being sold to compete head on with IBM.
Wah!? SWT is easier but otherwise no way superior. I haven’t used them much but I do know that much.. Java 7 is supposed to make Swing easier to use; if it comes out that is..
Who said anything about easier to use? I am talking about from a user perspective that Swing is an abortion of a widget kit. Sorry to fly off the handle but it is a horrible and disgusting widget kit that needs to be taken around the back of the wood shed and put out of its misery.
I want my applications to look AND feel like native applications; I’d love to see Java become a viable language for mainstream applications but it really tests my patience when I see there is no attempt to make a swing application behave like a Mac application. I never see this kind of problem when using SWT – and SWT will be upgraded from Carbon to Cocoa soon.
That doesn’t make it inferior but a difference of view — I say so because while the look is maintained; they’re often lacking in “feel” – but that MY PoV; and I’d sacrifice deceptive looks for a wholly unique L&F to avoid any confusion. I’ve rabidly avoided Java applications of the late anyway..
Or, use a ZVol, which gives you a block device with all the niceties of pooled storage, block checksums, and snapshots. Basically everything in the ZFS stack except the copy-on-write/transactional filesystem itself.
Does it protects against bit rot and silent corruption? If not, it is of no use in my opinion.
That I don’t know for certain. The end-to-end file checksums require the use of the ZFS filesystem. But ZVols do block-level checksums. So it should be able to detect some errors, and if you use redundant vdevs (mirror, raidz1, raidz2) it should (in theory) be able to fix them.
I haven’t played with zvols much as yet. That’s next in our test lab … using zvols exported via iscsi.
Awesome, thanks for the heads up – I’m surprised that Sun hasn’t made more noise about this and some of the features of ZFS. Sun always seems to be one of those companies who are insecure when it comes boasting of their own abilities – maybe the buy out by Oracle will give them some self esteem so that they can actually go out and trumpet some of the interesting features in Solaris.
http://blogs.sun.com/scottdickson/entry/fun_with_zvols_-_ufs
http://blogs.sun.com/chrisg/entry/zfs_pool_in_an_iscsi
http://blogs.sun.com/pgdh/entry/taking_ufs_new_places_safely
http://www.sun.com/software/solaris/ds/zfs.jsp
“Compatibility
Applications do not need to be changed or modified to use Solaris ZFS and take advantage of its industry-leading capabilities. It employs familiar POSIX interfaces, and existing storage infrastructure — device drivers, storage fabric, and devices — work without requiring changes. For applications that prefer to operate directly on block devices, Solaris ZFS provides the Zvol volume emulator, which delivers all of the benefits of transactional integrity and checksums, and is compatible with existing block-based, volume manager interfaces.“
Ask yourself – where are they located? you and I might go to the Sun blogs but are the people who make the decisions going to be visiting the same things you and I do? This is the one thing I have berated Sun over for years – their lack of realising that advertising in IT oriented magazines aren’t going to get the managers who make the decisions interested in your technology.
You need to get into management magazines like NZIM – heck, and if it involves having to have a article/marketing blurb of a customer who moved to your technology – and an article in the words of the manager for managers. Advertising on blogs that none of the decision makers in a larger organisation is an exercise in futility.
Can’t argue with you about that. Hopefully Oracle has a larger Advertising budget and can get the word out.
There’s not a lot of obvious press about a lot of the stuff in ZFS, but there’s a lot of info in the SUN employee blogs, and the ZFS Admin manual.
You can use zvols pretty much like any other block device. Use it for swap. Format it with UFS. Format it with some other OS. Export it via iscsi. Share it via CIFS or NFS.
You can even do thin-provisioning (create a zvol with quota size of 100 GB but with a reservation of only 10 GB) where the OS sees a 100 GB disk, but only 10 GB of real disk space is in use. If the app/OS needs more than 10 GB, just bump the reservation up, and more real disk space is available — but you don’t have to change anything in the OS/app, as it still sees a 100 GB disk. Very handy for storage with virtualisation.
All with the backing of the pool, including any and all redundancy features.
The only “big-storage” thing that ZFS doesn’t handle is automatic de-duplication (if there are 10 copies of a 100 MB file in 10 directories, only store 1 copy with pointers to the others, for example). But, inline compression mitigates this a bit. NetApp includes that in all their storage products now.
So you’re saying that Oracle bought Sun just because they’re only slightly interested in some products or services that Sun offers? Yeah right.
MySQL – Valuable to Oracle for it’s installed base, for Oracle sales staff to upsell as need arises. Also, gives Oracle entry into low to mid market. I see MySQl staying around, kind of becoming to Oracle what Access is to Microsoft.
Sparc – Will probably stay around for some very high end systems running the Solaris/Oracld DB combo, as Larry already touted, becoming an industry tailored appliance.
OpenOffice – might have value to Oracle, if for nothing else to tweak Microsoft (something Larry relishies). I do think Ellison sees the value in getting client business in order to help spur server side business. At least they’ll either breathe new ligth into it, or they’ll completely release it to the open source community.
Java – Major part of Oracle’s business (Middleware). But that’s just it – Oracle’s interest is in enterprise middleware, not on client side Java. So things like JSF, EJB3, and JPA have a very bright future indeed. But Swing, et al, will be there for legacy support, and developer tools only.
Did Steve Ballmer actually say, “I just learnt it?”
Maybe, maybe not. In any event, it’s not incorrect: http://www.askoxford.com/asktheexperts/faq/aboutspelling/learnt?vie…
Besides, it seems likely that the European spelling was used simply because the Reuters story came out of Moscow.
Edited 2009-04-21 18:32 UTC
I don’t know what will happen to everything else, but MySQL as we knew it has died and will continue to be driven into the ground by Oracle.
I hope everyone has moved to Postgres or one of the MySQL forks that have occurred.
don’t know what will happen to everything else, but MySQL as we knew it has died and will continue to be driven into the ground by Oracle.
I hope everyone has moved to Postgres or one of the MySQL forks that have occurred.
I don’t understand you doomsday proclamatists. Do you think MySQL will stop existing suddenly? It’s open-source, it’ll continue to exist as long as people wish for it to exist. Or is it because you think the development will stop if there is no big company sponsoring the devs? I wonder why there are so many projects out there doing just fine without such… Oh, and wait, there’s more: Oracle has not yet said anything even remotely in that direction. You’re just pulling that stuff out of your arse.
I’d much rather wait and see what actually becomes of MySQL. Besides, no matter what, it works just fine for me as it is. Even if things went all fubar’d I could still continue using whatever version I have now..
Have a look at the comment title, which is why I worded it like that ;-).
The exact opposite. It was better when there was no big company behind MySQL and they knew what they were about.
Considering how badly MySQL has been managed and led by Sun, and with all the original MySQL people jumping ship to create forks (if you keep up on news that is), do you think it’s going to get any better with Oracle who doesn’t want their expensive database products affected it? It’s not going to get any better, that’s for sure, and it seems logical that Oracle will even try and turn it into an ‘Oracle Lite’ database system.
You know, that’s why we look at what’s happening, look at what a company sells, ask ourselves whether it all logically fits and comment and discuss rather than waiting for someone to give us an ‘official’ line. Goodness me.
We’ve already seen the future:
http://www.itwire.com/content/view/23461/1154/
Odds are it ain’t going to get better for the ‘official’ Sun version. All we have to do is wait and see what the generally best and stable fork will be for future use when the dust settles, or look at something else.
Edited 2009-04-21 09:01 UTC
The more I think about it the more this is a total blunder by Oracle.
There is just NOTHING of value for Oracle here.
1) StorageTek…. a 2nd rate storage company that sun could not even market their products. Sun over paid for the company to begin with, laid them all all, discontinued most of their products and pretty much paid millions for a few JBODs which they had already.
Larry already has a storage company he started, Pillar Data which has products better than anything StorageTek had to begin with, and even Pillar is now learning that the days of big storage like their big arrays and EMC, Hitachi are coming to a close and commidity hardware is starting to take over the storage world in this decade like it did the server market last decade. Not much future in proprietary storage is a dead end product line as well, so that really does not make much business sense.
2) MySQL. Anyone running MySQL is not going to be ‘upsold’ into Oracle. Hell, I work for a fortune 50 that is almost 99.9% Oracle, spends millions of dollars every year in maintenance with Oracle, and EVEN WE are looking to REPLACE ORACLE as much as possible with MySQL over the next few years, NOT the other way around.
Maybe someone was under the foolish idea they could buy MySQL and destroy it to take its market share.. Bzztt..wrong that old business technique does not work in OpenSource. But I bet you $20 that’s what the business guys at Oracle thought they were buying.
3) Solaris/Sparc. Look, this is a totally dead product, no one wants to use it today compared to Linux.
And who in their right mind would want to pay for all the staff to develop, debug, maintain, support, a full operating system when all your competitors are all doing it for about 1000th the cost using the Linux / Open Source model.
Your NEVER going to be able to convince anyone with a clue again that a proprietary hardware platform or OS is going to compete with commodity Intel Hardware and Commodity OS’s like Windows and Linux. That 1990’s argument is not going to fly in the world of VMWare and Intel blade systems etc.
Maybe ZFS is interesting…but contrary to the hype ZFS is not all that, its just a good volume manager and filesystem rolled into one.
The real test will be to see if Oracle is more friendly to open source and make ZFS GPL2 compatible or if they will try to play the proprietary down with the sinking solaris ship that Sun played.
4) Open Office. I don’t know, maybe Oracle could do something with this, if they did a hard sell to their existing customer base and some back room deals, lots of expensive dinners to CEO’s to get them to dump Microsoft etc… probably just going to end up going its own way into a pure Open Source project that would have no problem getting sponsor ship from the open source community of Redhat, IBM, Novell/Suse etc.
5) JAVA. This is pretty much IT as far as something that Oracle can get any value out of. But, even Java is now fully open source, and starting this year with openjava being 100% compatible, Linux users have no need for SUN’s java anymore. I bet the business guys at Oracle thought they could purchase Java and control the Java market. Boy are they in for a rude awakening. The best asset here is probably all the great Java developers from Sun that could help accelerate some Java developments at Oracle since they are putting Java into all their products.
So…after reflection, I would bet that Oracle probably THOUGHT they were buying total control of Java, buying MySQL so they could take its marketshare, and were getting OpenOffice to maybe do something with….but those are all totally false assumptions due to the open source model not working like that.
What I see is that the REALITY IS it looks to me like Oracle purchased a bunch of pretty good Java developers talent for $6 billion dollars.
I am guessing we have a big case of buyers remorse coming once the board realizes they got taken.
I also know as someone with a fair amount of technical influence at a fortune 50, I will be watching with an eagle eye how Oracle behaves themselves with regards to these open source projects and the open source community. Their behavior could change my software recommendations, the product mix I choose to champion in designs, and what I talk negatively/positively about publicly to all the top level IT leadership in our organization when its time to make IT decisions.
Edited 2009-04-21 02:17 UTC
May be you can explain why majority of Sun’s hardware revenue is still SPARC hardware then. Oracle gets a $4/yr billion hardware business.
Also explain why if ZFS isn’t all that great Oracle is investing in BTRFS because of it.
Edited 2009-04-21 04:36 UTC
I’m not entirely sure that you have a complete grasp of open source and business.
There are a lot of open source projects where the author(s) and copyright holders are a small, contactable group, or sometimes a single individual. This group would typically publish their code under a copyleft license, which says that anyone else may use it, study it and even modify it for their own purposes, but if they re-distribute it then it must be done so under the same copyleft license. Doing anything else would require a separate grant of permission from the author(s) and copyright holders.
OK … that does not preclude the author(s) and copyright holders offering a software house business another separate paid-for commercial license for the same code. If IBM wanted to use, say, MySQL in an IBM commercial (closed-source) product, they would have to go to the MySQL author(s) and copyright holders and negotiate a proprietary license … IBM couldn’t do this under the copyleft license. This is known as “dual-licensing”. Different terms in the different licenses for the same code, different costs apply.
So anyway, now that Oracle have bought Sun, it is Oracle who becomes the copyright holder of MySQL. The original authors are no longer the copyright holders … that set of rights in MySQL is what they sold. This has two effects:
(1) Anyone can still fork MySQL and re-release the forked code under a different name as copyleft-licensed code. But they cannot now sell (to a company such as IBM) a separate commercial license for it, as that right has now been bought by Oracle.
(2) IBM (or anyone esle) now has to go to Oracle in order to get a commercial license to use MySQL in a closed-source product.
(3) Oracle are under no obligation to sell anyone a commercial license for MySQL.
This would all apply to OpenOffice, but perhaps not to Java. The reason for the latter is that their may exist other compliant implementations of Java that do not incorporate Sun’s copyrighted Java code. IcedTea is one such project which may not fall now come under Oracle’s control.
http://en.wikipedia.org/wiki/IcedTea
Edited 2009-04-21 04:58 UTC
Just to illustrate, here is a project which is a fork of MySQL:
http://askmonty.org/wiki/index.php/MariaDB
MariaDB cannot use the “MySQL” name, even if it were the exact same code as MySQL (which it isn’t). Trademark law prevents this. Oracle now owns the MySQL trademark.
AFAIK, MariaDB cannot sell a commercial license for MariaDB to anyone, since it is a derivative work of MySQL (since it still includes a lot of MySQL source code). Copyright law prevents this. Oracle now owns the MySQL copyright. For a commercial license, one must go to Oracle, and get the “real McCoy” MySQL. Oracle can charge whatever they please for that, and Oracle are under no obligation to sell such a commercial license if they are not inclined to do so.
However AFAIK, under the GPL license, the MariaDB project can release new versions, as long as they too are licensed under the GPL. Copyright law, and the original GPL license of MySQL, enables this.
Anyone wishing to use MySQL functionality on a GPL project of their own can currently get a version of MySQL from Oracle, or a version of MariaDB, and use it. AFAIK Oracle have the right to withdraw such open source permissions for MySQL in the fututre, if they deem to do so. MariaDB cannot do the same for their code, MariaDB must remain under the GPL.
I hope this is clear. If you follow all this, you then perhaps understand what Oracle have bought, and what they have not bought, when they purchased some open source projects.
Larry E proclaimed as he sees Solaris as the best Operating System out there for their needs. In the press release, they talk about two strong points; Java and Solaris.
I think they will tailor Solaris to Oracle. And run Oracle on ZFS. ZFS protects against silent corruption which no other file system does – ZFS is “just another filesystem”. It has unique features. All the other things beside silent corruption, is just icing on the cake. Silent corruption protection is the main and unique thing about ZFS. It is a big problem and the problem will grow even more. Read this to see why ZFS is so hyped (it is only about silent corruption)
http://queue.acm.org/detail.cfm?id=1317400
And for the new ROCK cpu, it has potential to be much much faster than any CPU. Just as Niagara, the lowend CPU, has several world records today will ROCK, the highend CPU, break several more records. Potentially in best case, ROCK allows MySQL on one machine to outperform a whole Oracle cluster, if all goals with ROCK are fulfilled:
http://www.theregister.co.uk/2008/07/28/sun_dziuba_tm/
I suspect MySQL will live and Oracle will make it very easy to upgrade to Oracle DB. And ROCK will be sold with Oracle DB.
ROCK server + Solaris + ZFS + Oracle DB + Java + JVM, everything owned and controlled by Oracle.
Actually you have that backwards. SPARC is quite open*1, and Intel x86 is proprietary as it requires a license from Intel or another proper licensee to manufacture*2.
*1
http://sparc.org/aboutOverview.html
http://en.wikipedia.org/wiki/SPARC
*2
http://findarticles.com/p/articles/mi_m0CGN/is_1999_July_1/ai_55048…
http://www.engadget.com/2009/03/16/intel-threatens-amd-with-termina…
You need to quit talking about “StorageTek”. StorageTek is a brand now, and only a brand of part of Sun’s Data Storage products. You also need to get a grasp of enterprise reality. Storage demands are growing at a phenomenal rate, and companies with good storage lines are going to see huge growth. Yes, Oracle might want to jettison the tape products, but the storage arrays, the new Unified Storage Systems, and the Thumper Storage Server can be huge products with the right marketing. Also, the Magnum (Sun 3456 Datacenter Switch) and the 3×24 switch could be very successful products in the Infiniband space, except hardly anyone has ever heard of them (after Schwartz’s one blog about the Magnum).
Working for a Fortune 50 company, you should know that your company loves to have support contracts so that it can at least claim to have someone to fix problems when they show up. Your Fortune 50 company is not going to go use the OpenSource version of MySQL without a support contract. If Oracle can keep you from going away completely, by supplying you a support contract for MySQL, they still have a revenue stream and a foot in the door for the databases that need more horsepower than MySQL can provide. If you’re already using an Oracle product, you are less likely to go talk to IBM about DB2, especially if Oracle can provide you a unified management system across all levels.
Awful healthy looking corpse, if you ask me. Since you combined two different things here, let’s break them up.
Solaris: open source operating system that Oracle will have complete control over. They will be able to push the database down deeper and deeper into the OS and create the absolute best linkage possible. They’ve been wanting to turn the database into an OS, you can see that from the direct hardware access and the HA that they’ve been building into ASM, OCFS2, and RAC. Now, they’ve got that basis, and its a basis that is well known and still heavily used in the marketplace. Solaris may have been loosing marketshare, but it is still huge. As for who would want to support their own OS… how about IBM (several), HP (a couple), MicroSoft (too many), as well as dozens of other companies that support their own OSes.
SPARC: Traditional SPARC architecture is not very good in comparison to commodity, but the T-series and the upcoming RK-series are designed for high throughput, which is perfect for the Oracle workload, specifically horizontal scalability with RAC. What a great architecture for Oracle to base an appliance on.
Quick, someone tell IBM and HP, who have both proprietary hardware and OS platforms. Tell Apple, who although they have moved to some commodity hardware are still charging a premium, and getting it on top of their proprietary OS. Which strangely, the Linux crowd seems to love….
ZFS is revolutionary. And, combined with existing Oracle technologies, such as ASM and OCFS, can be even more revolutionary. Tie that with RAC, and database clusters suddenly become easy and great.
Ah, I see you’re a GNU person. Sun (and ZFS) is already very “open source” friendly, they just aren’t particularly GNU friendly. There is more to open source than the GPL.
The wonderful thing about owning a technology, even if it is open sourced, is that you can control its market. If you control it in such a way that people really don’t like it, then with opensource, the market will go away and become “technology-NG”, but if you aren’t completely onerous, you’ll keep your market, no matter what opensource is out there. Oracle will own “Java” and will be able to dictate what “Java” becomes. If you don’t like it, you can create your own JVM or your own language, but you can’t call it Java, and if you want the world to use it, you have to convince them to leave “Java” and use something else.
I think there are new opportunities for all projects.
They can let the opensource programs continue to evolve mostly like the maintainers want, but create the ‘missing link’ to make everything work fluently with each other. Everything at a low cost for the companies.
Don’t forget they can make those projects OS-independant, so companies can still choose what OS to use, although Oracle could deliver a good working optimized OS for each purpose again at a very low cost. A dedicated one for the Oracle DB itself, one optimized for mySQL, webserver (also internal) and for the end users.
They might not gain a lot from the products themselves, but there is also the support for them.
But in the end it all comes to one thing, what does Oracle want to do with them. It can also mean the end of what these programs are now, but I don’t think (and hope) so.
Mysql makes a great club to beat MSSQL with. That’s where Larry’s going with this. Free MySql kills MS revenue on the low end where Oracle doesn’t compete anyway. Paid MySql hurts MS revenue in the middle, where Oracle was at a disadvantage. Having MySql in-house gives Oracle an inside when successful businesses move from middle-end to high-end.
Depending on how much “I hate MSFT” is still a driver at Oracle OO can serve the same function as well.
Owning MySql gives them the opportunity to make it more Oracle-like. Expect to see sequences soon.
I would not say MySQL is a competitor to MSSQL, MSSQL is too expensive to be that and too advanced, while MYSQL is free but shoddy,MSSQL, after all they started with a Sybase core and improved it marginally over the years. The real competition to MSSQL in my opinion is Postgres which from an engine perspective is way better (but harder to handle although the admin tools got way better over the years)
and does cost nothing and plays from a data perspective in a similar league with better performance and reliability if configured correctly!
1. If MySQL is indeed killed somehow; what’s lacking in postgresql that should spell doom?
2. What *exactly* is gonna happen about the filesystems? BTRFS is a direct contender to ZFS so would they put a stop to their filesystem; or maybe two filesystems which I find redundant