Submitted by MySQL AB and Sun have announced that MySQL has been bought by Sun. “Sun Microsystems today announced it has entered into a definitive agreement to acquire MySQL AB, an open source icon and developer of one of the world’s fastest growing open source databases for approximately USD 1 billion in total consideration. The acquisition accelerates Sun’s position in enterprise IT to now include the USD 15 billion database market. Today’s announcement reaffirms Sun’s position as the leading provider of platforms for the Web economy and its role as the largest commercial open source contributor.” More here.
So they intend to (try to) do something about the Oracle + Linux momentum.
I don’t know much about much, but I think this news is kind of a big deal
Looks to me like Sun wants a peace of the LAMP pie and realized that having excellent support for PostgreSQL is not enough.
I don’t really think this will make a large inroad on x86 web servers running Linux but for any Sparc houses who don’t necessarily want to struggle with the Sun Java Enterprise System stack, it could be a god send.
Only thing I’d be worried about would be putting all my support options in one basket.
If Sun couldn’t make excellent support for PostgreSQL work for them, how are they going to make support for MySQL work any better?
I guess owning MySQL will force people to deal with them whereas excellent PostgreSQL support doesn’t, but Sun’s always seemed a bit like Novell to me in that they are both skilled at killing technologies they buy.
On a side note, PostgreSQL is an outstanding SQL server, in my opinion, and more people ought to take a look at it.
Yeah I hope that really doesn’t happen. Like the last time the took up GNOME made JDS and then …
What? You got something against Gnome 2.6? Of course… they discontinued the Linux version.
Do we have four year old releases of MySQL^WJava Database Server and flagging Linux support in our future? I’m glad I put my bets on PostgreSQL.
I think it is more the realisation that sure, they could spend millions of dollars trying to push Postgres and gain marketshare – but the market is pretty much happy already with MySQL, so it makes little sense trying to go against the grain simply for the sake of doing so.
Its a good buy but personally if I was Sun, it would have been nicer if they had more cash on tap because Sybase would have been a wonderful product to put up against Oracle and DB2; throw it into their $100 per employee per year Solaris Enterprise System package.
With that being said, hopefully Sun will have the resources to be able to allocate towards turning MySQL into an enterprise level database, and improve the ease of use so that it can scale down to medium/small businesses who want an easy to use centralised database.
to improve Java acceptance on web hosters.
Sun often has great ideas. But imho they are not able to make it attractive for many people.
Example: JavaFX. Nice idea and I like the language. But there are no released tools for productive use. Not enough marketing etc.
Same with Java for web applications: It is ok for big companies, but why don’t they try to improve its acceptance for hobby programmers? They should try to make it easier and we need much more java hosters for hobby programmers.
I hope that the mysql deal is not the same as usual. Else Sun is able to pull down MySQL usage and spreading.
They obviously think: Hm we do it only halfway through, the rest will be finished without us. They have to make it more comfortable for developers and newbies to spread the technologies.
Greetings
Mike
Because almost every administrative or developer app written in Java is trashy looking and limited in OS integration.
Ever used the Citrix admin console? It’s a nightmare, right click doesn’t work as expected, you can’t type and watch the cursor and focus go to where your highlight has moved. If you remember Novell Netware, the nwadmin.exe was awesome, but their own java-based Console One was ugly and weird. I’m not even touching on eclipse.
I know it doesn’t seem to make sense, but I think Java would make headway if there were more native interface apps.
Those considerations are true, but irrelevant for server-side Java. I wish setting up a Java web app environment was as easy and accessible as setting up a LAMP server.
Whats wrong with eclipse? I like it both on Ubuntu and Windows (XP and Vista)
I don’t know, I rather like(d) Azureus (don’t get me started on Vuze), and think Eclipse works as expected on both Windows and Ubuntu (they feel native on both). Similar arguments are made about other cross platform GUI toolkits, like GTK (Pidgin – feels native on Windows and of course is native on Ubuntu), and some of Mono. I didn’t even know that Banshee and F-Spot were running on Mono at first – both very nice apps – especially F-Spot.
Sun does need to figure out how to help their developers choose the appropriate tool kits though. These good apps do seem to be the exception rather than the rule.
>I didn’t even know that Banshee and F-Spot were running on Mono at first – both very nice apps – especially F-Spot.
That’s because they use Gtk#.
This is not a question about the platform/language but about the toolkit. If you would use java-gnome instead of the other java toolkits your java app would look native on GNOME too and the same is true for the Qt Java bindings and KDE.
That’s because Azureus and Eclipse use SWT, which is the first Java toolkit that doesn’t look like total ass. It’s also rather nice to work with.
Swing looks nice these days, too. In fact, I’m now preferring Swing apps to SWT apps, in terms of look and feel and performance. Sun has made great strides in improving Swing appearance and performance over the last two Java SE releases.
But that’s off topic.
I’m wondering how this acquisition is going to affect Sun’s relationship with Oracle. Buying MySql puts Sun in direct competition with Oracle DB (at least at the lower end). And the Oracle/Solaris combo is a common and effective match.
Remember what happened after Red Hat acquired JBoss, putting Red Hat in direct competition with Oracle Fusion middleware (J2EE app server and tools)?
Oracle released it’s own copy of Red Hat.
Larry is not a good choice to have as an enemy.
Which went nowhere due to RedHat’s excellent reputation for support and Oracle’s exceedingly poor reputation on same.[1]
Larry makes a lot of mistakes through arrogance and overconfidence.
[1] http://tinyurl.com/2bkmxs
Edited 2008-01-16 18:13 UTC
ConsoleOne was merely half-finished. Its javaness had less to do with its unsuitability than Novell’s lack of focus.
NWAdmin did everything, except some things you did from the console. ConsoleOne, intended to replace nwadmin, did some things, but not the things done from iManager, and some things done from the console, or some things still easiest to do in nwadmin. iManager didn’t do much, except for the areas where it did everything.
Novell kept starting towards making a universal, standard config tool, stopping half way and starting again.
I firmly believe that c1 was the right way to go, if only they would *finished* the damn thing. You can pine for nwadmin all you want but the reality is that it had too many weirdly-behaving widgets to really feel native and was never very stable.
You are absolutely right about developer tools. It doesn’t matter how good your technology is, without good dev tools it won’t get used.
I hear you…You can quickly identify a Java application: Huge RAM usage, poor OS visual integration, slow. C++ is the way to go.
How can you relate the 2 languages? You could add more credibility to your comment if you were to compare Java with Python and said that python is the way to go. But C++? Come on now. They fill two completely different needs.
1) Does this mean that MySQL will stop moving towards making more and more of it’s products either dual licensed or closed-source?
2) I understand that Sun is a major backer of Postgres. will this announcement mean that Sun’s Postgres work will suffer?
There is always EnterpriseDB(www.enterprisedb.com). Seems much more of Postgresql Supported than Sun anyway.
I wonder what other buyouts will happen in 2008.
MS will buy Linux!!!111
so that’s why they are going to ship windows 7 so early
Apparently Oracle has reached an agreement to buy BEA.
http://www.oracle.com/bea/index.html
So there you are.
MySQL is riddled with bugs and gotcha’s . Sun fix this crappy product. Make a backup system that is as reliable as RMAN and fully integrated into MySQL . This is why I keep turning people away from MySQL it just sucks in so many ways.
I’d be interested in what specific problems you are encountering. I am currently working on the migration of a Web search engine from SQL Server 2K to MySQL 5 for a very well known business search company here in the UK. Apart from the rawness of the tools and the relative immaturity of its stored procedure language compared to TSQL, it is proving very successful. Speed is excellent, and I have managed to increase search accuracy by a very great factor.
The GUI tools are not on a par with Mnaagement Studio, that’s a given. But as a database server, and making judicious use of correct table types, index types and server variables, and by optimising the Stored Procedures, it is proving to be a great platform. Much better than v4.
I have no particular OS/Server preference – I am an MCTS in SQL Server 2005 and will soon be a Certified MySQL Developer (next month if I can get some time off work!).
My issues with mysql are not about if it does Stored Procedures or what level of SQL compliance it has. I don’t even care how fast it is. My issues have to do with two simple things. Backups / Restores and Snapshots.
When users had 20meg databases and they had the time to spare for a read lock for the dump to work. This was not a big deal. The issue becomes a horrid mess when you have 160G of data and you need to dump the data twice a day . A plain old SQL dump of this size eats Disk I/O like its going out of style. Now say I have Dumps / database / table , and I need to restore a dump to a new server. Say I have a 50G table to restore to a new box , a 50G dump is going to take a long time to read into a new server.
Say I just need backups once a day there is no database snapshot system. This has to be done on the filesystem with a read lock on the database. So on most UNIX OS’s this can be done but now that I have a snapshot of the datadir how do I say restore one table from an innodb database ? Yes I am aware of innodb files / table , but this is adding up to a pile of band-aids added to MySQL to make it work better.
When will they address this in a proper way say the way Oracle does this with RMAN .
See the post called “MySQL on ZFS” and its replies.
I’m sure they’re thinking about this.
There is a hot backup product for innodb. Not so for myisam. I’m not sure why you’d be using myisam if you’re doing a lot of updates, inserts, and deletes.
Using innodb and even without a hot backup tool, you could achieve something close to what you’re interested in with a combination of traditional cold full backups and binary logs. You could also use replication and backup your replica to achieve high uptime.
Edited 2008-01-16 20:57 UTC
Again this sort of stuff needs to looked at by SUN / MySQL . Innodb is nice but its going away. Filesystem Snapshots work well in some cases but not all. A Database which has the snapshotting built is better. A MySQL system that has non blocking backups, that work correctly, fast and consistently on various architectures, would rock. And yes both things can be with software already out there. I know I am asking for allot but Have you seen Oracle 11G RAC with ASM.
MySQL is releasing their own transactional data store with mysql 6, Falcon, so it’s not like there’s going to be a decrease in capabilities.
Given that Oracle has been developed for almost 30 years and is supported by a mutli-billion dollar company which employs tens of thousands, I would hope they would have something to show for it. And they do, but not everyone needs (or can afford) the capabilities which Oracle offers.
mysqlhotcopy?
http://dev.mysql.com/doc/refman/5.0/en/mysqlhotcopy.html
MySQL locks the tables, preventing any sort of modification, so that’s really a warm backup. A hot backup does not prevent table modification.
mysqldump can do hot backups, but in the thread we were discussing binary backups, so I didn’t make note of it.
Edited 2008-01-17 00:33 UTC
Anyone use Firebird ?
“You can’t use MySQL because we can’t buy support from someone in our Gartner quadrant.”
Finally. Yay!
My company is a MySql shop from top-to-bottom, this is a great news, however a little weary since we are running MySql on top of Linux. Hopefully Sun will iron the bugs in MySql, create an enterprise worthy console and just add more value to MySql.
I hope this means something cool down the road like MySQL running on top of ZFS’s transactional storage system.
I read an article recently in acm’s queue magazine about ZFS and both Bill Moore and David Brown commented about such a thing.
Companies don’t just buy other companies on a whim. I bet they have been thinking about doing just that for a while. I can’t wait.
Edited 2008-01-16 17:52 UTC
Oh god yes! Freeze the database, create a ZFS snapshot, release the database, pull a backup from the snapshot.
Don’t people already do this with LVM2?
I don’t know anything about LVM2, but I’d figure that those aren’t el-cheapo snapshots. Remember, ZFS has no-cost snapshots thanks to COW. Creating a snapshot there takes less than a second and the only diskspace used is are the changed filesystem blocks.
Edited 2008-01-16 18:41 UTC
No snapshots are “no cost”. Modified data blocks must be copied once they’re changed, and that takes up space.
LVM2 snapshots work the exact same way, they can be created instantly and are copy-on-write too.
\
Possibly, but as someone else replied to this they are “free” in ZFS. You can do a snapshot on every commit. Think of Apple’s Time Machine but for databases….rewind, fast forward etc.
Besides the snapshots you are removing redundancy.
A lot of database code is data-integrity stuff, transaction engines etc. All of this is now provided by ZFS.
Imagine how much simpler the database would be if it could trust the file system.
Of course they wouldn’t make ZFS required but it would make sense to be able to turn off all of the overlapping stuff.
Well, that’s what I thought the originator of this thread was after, and it sounds interesting. But the post I was replying to made it sound like you had to take the database off line to do a backup unless you have ZFS and MySQL-specific support for its features.
I believe original poster meant using ZFS object layer directly for database. The same way Lustre uses ZFS DMU directly for implementing higher levels of filesystem.
ZPL (ZFS Posix Layer) is such use also — it implements files and directories over ZFS object layer.
I tend to be a bit nonplussed over the idea of writing software to use filesystem specific features. There are two things that can be beneficial to doing so. Either the fs can do the job faster and more efficiently, or it can save the programmer some hassle and make the code simpler. The problem is that usually the software can do things in a more efficient way that suits its own requirements better. And unless the author is willing to tie his code completely to one filesystem, he’s going to have to do that anyway to support the other filesystems. So in the end, he would end up with more code that is more complex than if he did not try to use the special features of the one filesystem. This is why I have never been particularly excited about Reiser4. If the whiz bang feature is not in the vfs and available to all filesystems, the real world advantages of using it quickly evaporate, even if Linux is the only target.
I guess if *Solaris were the only target, it might make more sense.
Edited 2008-01-17 14:03 UTC
Well No freeze and a snapshot would be best.
write a full fledged AHCI driver and a bluetooth stack.
Sun might be able to replace that crap database in Open Office.
Not a very easy task I would think. I assume you’re thinking of replacing OO Base’s built-in HSQL db engine with one of those in MySQL, and ideally also milking the codebase for the management tools. There are at least two problems with that.
1) MySQL’s existing licensing model probably will be hard to integrate with OOo’s; yes, there is always the mysql-community version, but think about the userbase. The community and enterprise versions are closely unified at the moment, but depending what Sun decides to do with the licensing, that could change dramatically.
2) Each db engine (except MyISAM perhaps) has its own licensing quirks. SAP still co-owns the MaxDB engine AFAIK, and I recently learned that Oracle owns the InnoDB engine (this worries me a bit as I just finished a big project using InnoDB). Who knows what will happen with these engines once the dust has settled?
I can only cross my fingers that Sun has considered all the angles here and know that they can make this move without hobbling the product, but I’m far from confident of this given their track record. As for what kind of licensing they will end up with, even more doubtful about this being a Good Thing.
Now what does this mean for PostgreSQL?
Can we hope to finally get UPDATE FROM statement in MySQL ? Will SQL language implementation in MySQL become more resembling to other databases ?
SUN will iron out the bugs and turn MySQL to a real Enterprise product, with Enterprise support.
MySQL has crappy documentation. SUN has quite good docs, compared to for instance Linux (the source code is the docs), see the docs for Solaris or Java.
No more worries how MySQL will evolve. MySQL quit offering the latest binaries for download. Why? What would the next step be? Charge for the binaries? Fork off an Enterprise version and close the code, when MySQL has a sufficient great market penetration? With SUNs total commitment to open source everything; Java, Solaris, ZFS, Dtrace, Zones, etc – this is no worry anymore. MySQL will remain open, and given away. SUN will straighten MySQL up.
We all benefit from this.
Wasn’t the original appeal of mysql due to the fact that it was an incredibly small and fast sql db engine specifically because it lacked some of the functionality (and inherent complexity) of the enterprise full-featured alternatives?
I’ve used mysql in the past, but won’t even pretend to be an authority on database admin. However, IIRC, from digging into OSS db solutions, mysql was recommended over postgresql for performance if you didn’t need all the functionality, no?
I guess I’m just wondering if Sun really thinks they’re going to be able to build it into something more. $1B seems a little exorbitant, considering one article I read indicated mysql AB brought in $10M of revenue last year, and had only managed to convert 1% of their userbase to contract services.
Are they really going to try and compete with the incumbents, or will they settle for mysql being a cost-effective alternative that can play into areas that Oracle and IBM would be priced out of? More of an MS-SQL competitor?
The JBoss acquisition by RH made some sense to diversify further into the application “core” of organizations, even at the cost of falling out of favor with companies like Oracle and IBM as a result. I’m not entirely convinced this makes sense for Sun in the same manner, but that’s purely armchair quarterbacking on my part, so take it with a grain of salt.
That’s why they sell, after trying to cash de-synced enterprise version which they weren’t able to sell.
Will Sun fix the MySQL known scaling problems? FreeBSD dev guys have discovered them, and reported them to MySQL AB.
What this move means to PostgreSQL?
You might be interested in James Gosling and others stuttering, prevaricating, and otherwise saying nothing with many words, in this video presentation:
http://www.linux.com/feature/124909
I particularly enjoy the relevant segment on terraforming Venus. (One can only assume that hair transplants figure into the unspoken significance of the transaction, somewhere.)
Looks like MYSQL AB’s mixed license business plan paid off for the founders. It’s like an echo from the dot bomb^Wcom days.
Edited 2008-01-17 21:14 UTC