I really wish it had been IBM. IBM's investment in Java, open source, free tools, etc... has historically been stellar. Oracle, not so much. I would have felt much better about Java's future if it were in IBMs hands.
Of the companies, Oracle is the only one who has ever made really LARGE acquisitions and made them work. It will be interesting to see what Oracle does when they need to start spinning off the different tools that duplicate their functionality. Oracle has been dev friendly with free tools and development environments. You can download their enterprise Oracle database from their website. IBM always seems to charge for their tools.
My experience has been the opposite. Yes, you can get Oracle XE on a limited set of platforms (still no OS X support for any current Oracle databases), for free, although the default configs leave a lot to be desired. However that doesn't compare to the large number of free and/or open source Java projects, libraries, tech articles, forums, etc... that IBM has supported. http://www.ibm.com/developerworks/java/.
You also need a registered account to do pretty much anything on Oracle's site, in stark contrast to IBM's developer tools and sites.
Also if Oracle's SQL Developer is them being friendly to devs, then I wish they wouldn't:) That thing is one of the ugliest, most prone to crash, DB tools I've ever used.
Which just came out on the 9th of April I think, which is how many years late? Also, it doesn't appear to install correctly unless you're running OS X Server. Fine for production, useless for doing development on my laptop with.
Eclipse, that was (one of) the best Java IDE when I was still in Javaland, was originally an IBM project that they've released as open-source and they've been a major contributor to it afterwards. So IBM doesn't always charge for their tools.
Thinking about these large acquisitions, it's hard to imagine that Oracle now owns Solaris, Java, BEA, Seibel, Peoplesoft, MySQL and of course their own Oracle DB.
Its now about "cost-free" tools. The open source community can produce those -- if anything, Eclipse/NetBeans could just be extended with more plugins.
Its about open specifications and standards. Sun had been hyping Java 7, but hasn't been going through the Java Community Process to "debug" the spec, and allow for OSS developers (like those at IBM) to develop spec-compliant tools of their own. If Oracle continues down that trend, it will see Java lose its most important asset: a community of developers who believe in open standards.
MySQL is a low-end complement to Oracle, but Oracle already had a whole range of products in the Oracle range, including a free version of Oracle (Oracle XE). If Oracle really wanted to give away a free database, they would raise the limits on Oracle XE to something practical.
It is silly to spend resources developing MySQL in parallel to Oracle. I expect that MySQL will slow MySQL development (it is already glacial), and provide migration tools from MySQL to Oracle. Perhaps they will develop a MySQL front-end to Oracle and/or enhance MySQL's front-end to be more compatible with Oracle's syntax. Either way, MySQL has always been an inferior good and Oracle will ensure it always will be. Oracle just needs to keep MySQL alive enough that nobody else can start a viable business around a fork of it.
I also don't think that Oracle will want to keep supporting both Linux and Solaris in the long run. Now Oracle has ZFS and BTFS; they don't need both. It will be interesting to see what they do with Solaris and its key technologies.
My hopes are on them finishing BTRFS -- it'd probably take a lot more time to butcher ZFS into something that could be shipped by Linus (It took years for XFS, and ZFS is way more proprietary about VFS shit).
Oracle doesn't need and probably doesn't want it to be shipped by Linus. They just need to license it under the GPL and convert it to the Linux filesystem API. Then they can ship their own Oracle Unbreakable Linux with ZFS. They don't need to send in patches to integrate it into the kernel officially. Oracle can maintain its own private source control for the ZFS kernel module, refuse any third-party contributions, and put tarballs on OTN when they actually ship releases. Officially it would be GPL'd but practically it would be useless to any vendor except them and their partners (RedHat).
They'd spend quite a bit of their social capital with the Linux community if they tried to be private that way.
I've also heard rumblings that a primary motivator behind the CDDL (and GPL incompatibility) was a core group of old Solaris devs at Sun that would have been royally pissed off if their work was GPLed. I wonder if Oracle would care if they quit en masse.
I don't think Oracle really believes there's value in "social capital". To the contrary, they seem to thrive on the fact that almost everybody despises them a little bit. If people were happy to do business with Oracle then Oracle would feel like they weren't charging enough.
Oracle didn't have to buy Sun to have these benefits. The only reason this fits better than an IBM acquisition is because Oracle and Sun internal cultures are more aligned.
There's benefits. First they can start making MySQL a gateway drug into Oracle. e.g. aligning syntax, a clear demarcation of features between the products, upgrade programs, documentation, support, etc.
In addition to the mysql question, I wonder what it will mean for other open source stuff, like Java, which has been moving in that direction, Open Office, Solaris, etc... Oracle, traditionally, hasn't been a big open sourcer of things, although they were one of the first to give some real legitimacy to Linux by porting to it, some 10+ years ago.
Also, culturally, how will the fit be? Wonder if they intend to keep some of the good bits of Sun and trash the rest or try and run it as a going concern?
"I wonder what it will mean for other open source stuff"
Looks like it's time for the open source model to prove its stripes. One of the main selling points for using software with an open source license is that you can get someone else to service/extend it for you even if the originator of the code goes out of business or shuts down development.
It will not be surprising if Oracle wants to take some of the Sun open technologies in a more proprietary direction or just show less interest in continuing development. That will mean we will see whether or not the purported benefits of an open source license prove themselves or not.
The main challenge will be forking, assuming Oracle continues to develop MySQL, for example, but only in ways that do not encroach further on Oracle's existing DB market. If people want features that Oracle does not want to provide, it looks like forking will be the only option.
> Looks like it's time for the open source model to prove its stripes.
Errr... it already has, over the last 10 to 20 years. That doesn't mean that projects don't go through ups and downs though, and Sun has, as of late, been contributing to stuff like JRuby. It won't die if they pull the plug on the paid developers, but it might be a setback for the project.
It's probably true that projects like JRuby/Jython/Scala/etc won't die (especially Jython what with BEA and all); I'm more concerned about things like invokedynamic and friends from JSR-292.
Hopefully Oracle will be friendly to Sun's research-like teams that may not be making direct contributions to the bottom line.
Have you ever been in a company during a merger? (I'm on #3 now.) There's always the potential for that. Internal power fights are common in times like that.
Oracle is written in two layers, there is an abstraction layer and Oracle itself. If you know Oracle then you know it handles a lot of stuff for itself that the OS would traditionally do, for example it has its own process scheduling, interprocess communication, filesystem/volume manager, memory management - as of 11g it even talks directly to NFS servers in userland, bypassing the OS filesystem mounts entirely. Once the abstraction layer is ported then yes, it would be just "make" for Oracle itself.
This is how back in the day Oracle ran on 90+ Unix variants...
Because of different shared memory optimizations in Solaris (especially Solaris 10+) and Linux, there's a non-trivial amount of work involved. I doubt it was anything close to "just type make".
One of the things Oracle does VERY well is acquire companies. They do it a lot more often than most people realize and have it down to a fine art. That does not mean that everyone fits into their culture, but they'll identify those people as part of their process.
They'll start by putting an Oracle logo on the Sun product and service range and keep selling them to exist Sun customers and then start on existing Oracle customers (where Oracle don't already compete).
Next step is to more clearly position, integrate and possibly rationalize the new offerings.
Oracle owns InnoDB and Sleepycat/BerkeleyDB, both of which are in active development. InnoDB is also MySQL's most popular storage engine. So I wouldn't fear for MySQL's future. It wasn't doing well in Sun anyways.
Now MySQL doesn't have to implement a new engine to replace InnoDB (their previous motivation for the new engine was to keep control of it away from Oracle).
MySQL and Oracle excel in different areas. I would expect Oracle to halt any effort to grow MySQL towards Oracle's feature set.
I suspect this is very bad news for Postgresql. Sun is a major backer, but that will likely end. Postgresql's target market already overlaps with the low-mid end of Oracle's, and Postgresql 8.x has been largely about scaling up.
Actually, it's interesting to note that of those 17 top contributors not a single company employs more that one top contributor.
If anything this is good news for PostgreSQL, then: Oracle can easily acquire and steer the course of MySQL development, but shifting PostgreSQL priorities by financial means would take the acquisition of Sun, IBM, UC Berkeley, and over a dozen more companies.
Even when Josh worked there I think you'd be hard-pressed to make an argument that Sun "backed" PostgreSQL in any strategic sense.
It's true that Josh is an extremely vital figure in PostgreSQL's community, but he doesn't contribute code as much as he advocates and interacts with the user community. Those roles are extremely important, but the people who hold them are pretty much beyond outside influence: Josh will always act in the best interests of PostgreSQL regardless of who signs his paychecks.
On top of that, these types of evangelists don't usually play a part in the technical direction of a project. I'm not sure exactly how closely involved Josh is personally with PostgreSQL's technical direction, but I can tell you for a fact that any attempts a company might make to steer a project through its evangelist are misguided.
Either way, though, there's no question that Josh is incredibly important to PostgreSQL. But, still, that's a whopping two contributors out of over two dozen.
We can go back and forth about what level constitutes "major backing", and that's fine. But at least now we're talking about some data instead of just hand-waving; that was my point in posting the initial correction.
Yes, one area where MySQL excels is it doesn't cost $40k per server to have an unlimited (e.g. wired up to the internet, so anyone can use) license.
PG said in one of his rants, "Do you know of any start ups choosing to go with Oracle? I don't either."
A whole new generation of companies (like RoR dev shops) are using Postgre/MySQL as clusters/shards and have no intention to ever moving to Oracle. This way, Oracle still captures revenue from them. (If they are paying for support).
Well, losing Sun as a backer would be bad for PostgreSQL, definitely, but as far as the future goes, for PostgreSQL the sky is still the limit, for MySQL the limit now seems to be Oracle.
Wouldn't other hardware vendors be more interested in supporting PostgreSQL now? Before the acquisition, if you bought a big HP database server, HP had no reason to care which database software you ran on it. Now, HP would prefer that you not be directing your software-purchase money to one of HP's competitors.
Because Oracle and HP are already tight partners and analysts refer to them as essentially one entity against IBM. Plenty of people expect the Sun hardware line to die or get sold to HP, while Oracle keeps the software line.
I would definitely be concerned about any future innovation of MySQL. Perhaps, it will get folded into Oracle's Express edition and then slowly phased out over time. Any advances that put MySql even remotely close to Oracle in terms of scalability are unlikely to occur.
Overall it might be much better than the deal with IBM. With IBM it was a forgone conclusion that Solaris would have vanished, in case of Oracle there is a fair chance it will survive(look at Weblogic - Oracle is promoting Weblogic as their primary middleware). With this purchase, Oracle is in a position to provide(and own) the whole stack : hardware/os/app container(java)/app/database - look around the only other company which comes close to this is Microsoft(which does not have a big ERP portfolio). I think MySQL was just a small factor in this deal. I think this will be big loss to IBM(now that Oracle gets it). IT customers are gravitating towards utility and end-to-end solutions and applications usually drive push(guess who has the biggest business application portfolio).
With IBM it was a forgone conclusion that Solaris would have vanished
IBM currently sells OS/390, OS/400 and AIX while also pushing Linux and only recently stopped selling OS/2. I think it's quite likely that Solaris would have just joined the list.
Solaris 2.5 and 2.5.1 were ported to IBM PReP systems in the 90s. You could order certain models of PPC workstation with AIX, Solaris, Windows NT, or OS/2 preloaded.
Was it all pre-planned? Sun buying MySQL at 1 billion just to make itself more attractive to Oracle for a buyout within a year?
In the end Sun seems to have made a profit out of its 1 billion investment. Oracle probably have added a lot than it would have offered simply to ensure it owns its main business killer MySQL.
I doubt Sun planned for this. If it hadn't been for the financial downturn, Sun's acquisition of MySQL (with its 12 million customer base) could have been Sun's path to new profitability. Maybe not a guarantee, but I suspect they were hoping for growth rather than a buyout.
The only thing that comes to my mind is: Someone's had to do it.
But it's still interesting that it's Oracle who did. This brings InnoDB/BDB and MySQL and of course Oracle under one umbrella. Also, Java now 'belongs to' Oracle, as does Solaris, xVM/VirtualBox, etc.
From my ignorance, it seems a very smart move from Oracle. It makes a lot of sense for them since Sun has lot of things they lack of: Server hardware, the Java stack, an operating system, their own processor line... They just became an IBM or HP!!
IBM already has all of that, and they sell the complete stack: zSeries/iSeries servers - Power Architecture - zOS/OS400 - DB2. Now Oracle may compete with Sun M9000 - SPARC - Solaris - Oracle DB. Amazing, isn't?.
I understand that people around here is worried about's MySQL's future and the like, but I believe that this acquisition is about the big bucks of the high end servers, and the fact that Sun owned MySQL had zero relevance.
This is interesting because I looked on Oracle's website for the deal expecting to at least see a percursory promise to keep things open but didn't find it. Who knows?
In the conference call they talked about operating Sun at a significantly higher margin, and talked only about Java/Solaris/Sparc/Storage + Oracle synergy. The sum of references to open source was, "oh yeah, and Oracle still runs on Linux".
I'm a bit of a pessimist, but I'd expect to see serious downscaling of FOSS at Sun.
"Will the ownership of Solaris change Oracle’s position on Linux?
No. This transaction enhances our commitment to open standards and choice. Oracle is as committed as ever to Linux and other platforms and will continue to support and enhance our strong industry partnerships."
Yup. They rejected IBM's offer for 9.40 a share, but are accepting this one for 9.50 a share. After their stock plummeted for rejecting IBM, I suspect they didn't want to pull another Yahoo.
This is all interesting, but I think IBM may regret not getting this deal done. It seems "owning" Java gives Oracle some leverage over IBM as so much of their services business hinges on Java. The possibility of turn key solution stacks with hardware/Solaris/Oracle/business apps might give Oracle a chance to compete in IBM's markets, too, or at least reduce demand for IBM services. If you can get all that installed and supported from one vendor, you might need fewer service agreements from third party vendors like IBM.
IBM lowered it's offer after discovering a few things about Sun it didn't like. Sun was OK with the price, but added some legal causes IBM didn't like - so IBM walked away.
Then there were also the regulatory issues that the IBM+Sun deal would have faced. IBM and Sun are the main competitors in a few areas (large scale tape storage, for example). This would have made the deal more expensive for IBM than just the $7 billion price tag.
Oracle and Sun have very few overlapping markets (I can't think of any off hand, aside from the Oracle Red Hat clone and Solaris). As such, this deal will be much smoother for both parties.
As far as their larger products are concerned, I think that Oracle is a much better fit for Sun than IBM was. Obviously there will be some overlaps in technologies (MySQL vs. Oracle being the big one) but on the whole, I think these companies match together pretty well.
Are there any other huge overlaps in technologies (other than the obvious MySQL/Oracle)?
Definitely not as big as MySQL/Oracle, but one that I'm wondering most about is their open-source app server Glassfish. A big part of Oracle's BEA acquisition was bolstering their middleware suite (with WebLogic, et al.). Will Glassfish development continue? I hope so...
We're on the scheme, and use Sun x86. Maybe it's different here in the UK, but it's been wonderfully pressure-free on the sales side and they've put an awful lot of effort into helping startups in ways other than selling cheap kit.
Obviously it's all out of self-interest, but it seems to me to be very much enlightened self-interest. Which isn't how you'd describe Oracle.
Odd, I'm UK (London) based too. I encountered their salesy SSE people at FOWA London. Politely had to explain that I has no need for 256 concurrent threads on a single machine in any of my web apps at the present time.
Maybe I'm just out of the loop but, this kind of struck me out of left field. All in all, I think this is aimed to keep Oracle competitive with Microsoft outside the database realm. Now, just like MS, they control the Operating System (Oracle's Linux is no match for Solaris), the language, and the database. Plus, they have the added bonus of controlling the hardware too!
"Now, just like MS, they control the Operating System (Oracle's Linux is no match for Solaris), "
Really? The telco and finance industries that were all Sun based in the mid 90s have dumped most of their Solaris for Linux. Startups are still predominantly Linux.
Initially they were a FreeBSD shop, I think Sun gave them heavy subsidies to switch to OpenSolaris on Sun hardware. I'm not sure how much the sweetheart deal was relaxed as Joyent grew.
I'm curious to know that too. Sun was pushing VirtualBox(with more features and killer performance improvements) in order to sell more machines and promote virtualization as the way to the future.
I am not sure if Oracle would actually be interested in this.
Less clear-cut than the future plans for the database servers are those for Sun's ZFS and Oracle's BtrFS. Are they likely to continue development in parallel, or might the licence on ZFS be changed and development merged?
It actually makes it marginally more likely (Oracle's not going to be so worried about pissing off old anti-GPL Solaris engineers), but the license is the least of the problem.
ZFS would have to be completely rewritten to ever make it into Linus's tree -- it's written for kernels that don't really have a VFS layer, so it implements its own, and the linux kernel devs are pretty anal about sharing implementations. Once you fixed all the VFS issues, you'd end up with essentially a BTRFS clone that uses ZFS's on-disk format.
Incidentally, BTRFS's development is sponsored wholly by Oracle!
I'm not sure it's fair to summarize the Linux community's complaint about ZFS as the latter being "written for kernels that don't really have a VFS layer." Solaris very much has a VFS layer, and ZFS plays with it. What's different about ZFS is that, from most storage stacks' perspectives, it is also a volume manager. Linux people claim this is a layering violation; ZFS' architects claim that integrating traditional functions of filesystems and volume managers let them innovate, e.g., by closing the RAID write hole.
The best news that could come out of this deal would be better Linux support for T1/2/2+ and Thumper hardware. Awesome hardware but I can hardly justify building expertise and support infrastructure for Solaris just to run it.
NOOOO! Sun has done many great things for the computer world, and I'd hate to see them be locked behind Oracle's doors so where only big corporate guys that smoke cigars and drink fine aged bourbon have access to them.
There will be a conference call today to discuss the transaction at 5:30 a.m. Pacific time. Investors can listen to the conference call by dialing (719) 234-7870, passcode 923645. A replay will be available for 24 hours after the call ends at (719) 884-8882, passcode: 923645. A live audio webcast of the call will be made available at www.oracle.com/investor and a replay will be available for seven days after the call ends.
