Blue Sun? What an IBM acquisition of Sun means for software

By Kurt Cagle
March 23, 2009 | Comments: 17

The recent posturings of IBM (and more recently Cisco) in acquiring Sun Microsystems has provided significant grist for financial analysts who look upon Sun as being in the main a hardware company. Certainly, the acquisition of the various sun server lines would help IBM, already the largest provider of server hardware in the world, with an even broader suite of offerings, especially on those specialized markets such as media production and hosting, universities and academia and higher end Internet farms.

However, Sun's software side of the acquisition ledger, especially if by IBM, has been rather oddly overlooked, given that it will likely have major implications for software development and cloud computing for years. Sun's software holdings cover five primary areas - Java, Solaris, mySQL, Open Office, and Sun's recently acquired QLayer cloud infrastructure. Understanding how IBM could potentially ramp up (or destroy) each of these gives some interesting insight into the real value of IBM's potential software acquisitions.

Java

Java has long been one of Sun's crown jewels, though even from the days when Java was called Oak Sun's had an awkward relationship with its star child. Java's growth has slowed dramatically from its inception in the mid-1990s, yet even after nearly fifteen years, Java continues to replace C++ as the tool of choice for universities, infrastructure development companies and software developers in general.

Sun, however, has been reluctant to completely open source Java, creating licenses on the language that still keep the evolution of the language firmly under Sun's control - and as such out of bounds for most Linux distributions to include in a purely GPL/LGPL stack. Moreover, it also firmly keeps the Java trademark and moniker fully available for Sun to use to promote its own products.

IBM surpassed Sun in 2007 as the company with the largest number of Java developers, and while IBM has made extensive use of the language internally, the tight reins that Sun had over the Java Community Process often meant that changes that IBM wanted to see incorporated into the language were slow to happen, and in many cases Sun priorities affected both which new class libraries, core langauge changes and modules were added and how those changes were implemented.

It is likely that with an IBM buyout, Java itself will almost certainly be made available in a GPL version, which will also have three rather interesting effects. The first is that it should prove a boon to Linux distro developers, who can include a whole raft of Java applications that have previously been limited in scope to "commercial" add-ons, and as importantly, it raises questions about the degree to which Microsoft can include a Java distribution in Windows (assuming that Java isn't dual licensed).

It will also turbo-charge the Java community, whose participants have drifted away from Java in recent years not only because other languages such as Ruby or Python are sexier, but because the Sun license has hobbled its use in the increasingly open-source oriented software development groups and projects.

Solaris

Solaris has run into the same licensing issues that Java has - a Unix kernel that straddled the boundaries between being open-source and proprietary, with none of the benefits of either. While this has meant that it has evolved a fairly sophisticated ecosystem of products, typically its also meant that Solaris versions of popular Linux tools would take a year or more before appearing.

Open Solaris, the open source version of the language, had been launched under Sun's CDDL (Common Development and Distribution License), which has been developed primarily to remove encumbered code which had exclusive Sun proprietary patents in 2007. The CDDL, however, is based upon the Mozilla license, and again raises licensing issues with Linux distributions because of the structure of the license itself. This has meant that Open Solaris code (or anything built under that license) could not be incorporated into a Linux distribution, making for increased balkanization between the two platforms.

IBM will most likely change the licensing on Open Solaris to a pure GPL one. This will have the dual effect of making it legally possible (technologically its quite possible) to run Linux applications under Open Solaris, significantly enriching its code base, while at the same time allowing reverse pollination in those areas where Solaris has a significant edge (such as the libPing stack).

There's some question about what will happen to the proprietary version of Solaris, however. It's possible that IBM may choose to continue Solaris development, but it's more probable that Solaris 10 may end up being the last of that proprietary branch, with any subsequent development (beyond big fixes) taking place exclusively on the OpenSolaris branch (and with Sun's current Solaris customers encouraged subtly or otherwise to switch to the OSS version).

mySQL

Sun's acquisition of mySQL in 2008 caused a number of analysts to scratch their heads trying to figure out what exactly Sun's strategy was in this regard. While mySQL is highly popular in the open source web stack, as an open source project (even with secondary commercial licenses) it was freely available to Sun to license for Solaris, it didn't add appreciably to revenue (indeed, it can be argued that Sun's purchase significantly degraded their own financial health) and it didn't even offer up that much of a consultant arm advantage, because it is a comparatively easy database to both setup and use. The mySQL database can be seen as a precursor to a cloud platform, but even there, it is a very minimal component in that particular domain.

If IBM acquires Sun, then it also acquires mySQL. However, for IBM, mySQL represents a very different value proposition. First, mySQL gives IBM a superb stepping stone to higher-end IBM DB2 functionality, and perhaps as importantly it has the potential to fuse the seat-of-the-pants mentality of young mySQL developers with some of the more traditionally staid DB2 developers, in all likelihood making DB2 as a database much more focused on the web developer (who increasingly also acts as database administrator at the mySQL level).

On the flip side, IBM also has the opportunity to ramp up funding to mySQL in a way that Sun was unable to do, dramatically accelerating its development cycles. This probably has more significance in the European market, where mySQL had a fairly significant presence already, than in the US, but it does represent an effective market for IBM, especially at a time when open source technologies may factor large in governmental programs, an area that IBM has considerably more at stake than Sun did.

Open Office

Sun has had a rather interesting symbiotic relationship with Open Office ever since it released the source code for Star Office in 2000. While Sun doesn't "directly" control open office, it does have an agreement to port code developed from Open Office into the commercial Star office suite, while augmenting Star Office with proprietary extensions.

It's noteworthy that during the most recent OOXML/ODF standardization scuffle within the ISO (which has in effect made both Open Office and Microsoft Office XML formats standard ones), that the most vociferous champion of Open Office was not Sun but IBM. IBM has a huge stake in establishing an alternative office suite to Microsoft Office, as this approach is very much in line with its open source consulting strategy that has proved quite effective.

Here it was not necessarily an issue of license so much as the more immediate need for cash that has slowed Open Office development over the last few years. At a certain level of maturity, any open source project needs full time developers and maintainers, and while there is an active Open Office community, it is almost certain that IBM has been the primary supplier of that particular largess, not Sun. Thus a buyout of Sun here would not necessarily change Open Office dramatically (though it would improve it somewhat), but would cut down much of the friction between IBM and Sun about how best to push. build and promote its features.

Cloud Platform: Q-Layer and OpenStorage

In 2005, Sun purchased StorageTek (ironically founded by four IBM engineers in 1969) and released a number of their products under CDDL licenses as the OpenStorage suite. The infrastructure for OpenStorage happens to contain many of the key hardware virtualization layers necessary to create Hardware Clouds, including virtual network stacks, virtual disks, and disk array controllers.

Sun's acquisition of Q-Layer in 2007 further strengthened this position. Q-Layer provided large-scale cloud management software, that, in conjunction with OpenStorage established much of the backbone of Sun's cloud strategy. This was also integrated with Sun Grid, which was one of the first formal grid-based computing systems (and by extension, cloud systems) in the market, as early as 2003, although the program gained relatively few adherents until comparatively recently.

IBM itself has been making significant forays into the particular realm of cloud computing for the last eighteen months or so, most notably with IBM Blue Cloud initiative. One of the most telling points about this initiative, however, is that it's not so much reflective of a single given technologies but instead consists of a fairly wide spectrum of alternative, albeit (more-or-less) compatible, related technologies.

Cloud strategies seem to require about eighteen months to two years to come to fruition, and as such there's considerable value to be gained for IBM in getting a hold of the Sun Cloud (as Sun's latest incarnation of distributed computation is called). What's more, IBM's Blue Cloud has suffered considerably from a perception standpoint - seeming somewhat nebulous and ill-defined (rather appropriate for a cloud, perhaps, but not for enterprise customers looking at significant financial outlays for virtual systems and services).

Sun's strength has long been in the Internet server market, so the jump to a broad cloud-based strategy is perfectly natural for them, whereas IBM's servers have typically been geared more towards a more classical enterprise strategy that doesn't necessarily mesh with the distributed point-to-point service architectures that seem to so typify cloud computing. Thus, acquisition of Sun Cloud could definitely serve to strengthen IBM's presence in the "Internet Enterprise" market, at what amounts to a bargain basement price.

Setting Sun

At least one pundit has raised the very valid point that, in trying to have its cake and eat it too with regard to Open Source software, Sun may have actually delivered a mortal blow to itself. Sun CEO Jonathon Schwartz put forth a compelling set of arguments recently about how Sun could monetize Open Source software. Unfortunately, this realization has likely come too late.

IBM took a different tack, more than a decade ago, of absorbing its open-source software development costs in order to gain effective control over providing open-source related consulting services. IBM's net revenue is down somewhat from last year, but overall it has held up well. Sun, on the other hand, has continued to spiral downward as its core products - Internet servers - have seen sales plummet, as ambiguous strategies and timidity in embracing a fully open source model has hurt key technologies such as Java, and as questionable business purchases such as mySQL have depleted its operating capital (and likely placed it at the whims of the currently brutal credit markets).

Recessions reward prudence and foresight and punish bad business decisions and excess. It's unfortunate that Sun, long a home for some of the most innovative products, services and ideas in the tech sector, is now faced with the consequences of those decisions, but its not, sadly, all that surprising. This should be a year for consolidation as cash rich companies go on a buying spree to pick up good companies at bargain rates, and Sun will be neither the first, nor the last, to be bought up or go out of business.

Kurt Cagle is an O'Reilly editor, and is the webmaster for XMLToday.org. You can follow him on Twitter.


You might also be interested in:

17 Comments

Do you have any idea what will happen with the Sun IAM products (Identity Manager, Access Manager, OpenSSO)?

java: IBM is more likely to balkanize the language, widely regarded as the current COBOL, as it did to COBOL itself when ANSI got into the act. It is already necessary to use IBM's jdk for some of IBM's java stuff, notably WebSphere. M$ could have done what IBM did, had they been smart enough to understand that OO coding allows you replace any functionality you want, and still stay within the Sun rules.

MySQL: the sui generis version has never been a *database*, merely a sql processor fronting the file system. With InnoDB, now owned by Oracle, it approached RDBMS. The "transactional engine" that they were building died when its developer bolted. In many real ways, the conversion of COBOL/VSAM applications on the z/OS machines would make more sense going to MySQL (were there a z/OS port) since not only would it be more honest, but would also be cheaper. Such conversions, and I have seen a few, seek only to dump the files into DB2; the relational facilities of DB2 are ignored. The goal of such conversions is to claim that the application is now a Database System, not to BE a Database System; the ACID is left in the application code. I was not rewarded those times I made the suggestion!! Such systems are what KiddieKoders do with MySQL.

IBM could call it DB2 for COBOL-z/OS. Bet you a nickel that's what happens. Perverse, but sly.

This article contains misleading statements. First, it implies that Sun is somehow resistant to the idea of open source, saying that they are reluctant to make Java completely open-source. This is not true; they've been quite clear about their dedication to open-source Java and have already released a preliminary GPL version of Java 6. Some Linux vendors are already including it in their distributions. The final release of Java 7 will also be GPLed.

Sun has many other open-source products as well:NFS, MySQL, OpenOffice, GlassFish, Netbeans, and more. Compare this to IBM: They've got Eclipse and...well, that's about it. If you want to cite a company that's resistant to open source, Sun should be the last on your list.

The article also states that IBM might change the licensing on OpenSolaris to a pure GPL one, making it legally possible to run Linux applications under OpenSolaris. This is a non-sequitur! There are no legal restrictions on running GPL applications on OpenSolaris, just as there are no legal restrictions on running GPL applications on a closed-source OS such as Windows.

Trevor,

My contention is not that Sun has not been pushing toward open source (and nor was my intent to mislead here). Rather, that it came to the strategy of moving to GPL too late, especially with regard to both OpenSolaris and Java, to save it. Had they pursued this strategy with regard to Java in particular (and given up their tight hold on the Java community process) four or five years ago, it is very likely that they would be in a much better position now.

It has been my impression that OpenSolaris was also on the CDDL, and I've seen nothing to counter this. There's no restrictions about running Linux programs on OpenSolaris (which is not what I meant), but there are definite restrictions about being able to incorporate Solaris-based libraries into Linux applications because of the same CDDL, and I believe the same restriction works the other way as well.

In essence, the crux of my argument is that the CDDL, intended as a way for Sun to retain control over their open-source products, has instead worked against them. Personally, I think Sun is a superb company with some of the brightest, most innovative people in the industry, and in many ways I'm sorry to see them reach the state where they could be bought up so cheaply, given the assets that they do have. This article has just been looking at the consequences of that fact.

Jan,

I'm not sure, to be honest. I would guess that many of the second-tier tools that Sun provides will likely continue in their present forms with IBM re-branding, at least for a while, but a lot of that will depend upon the specific arrangements that are made with regard to those tools.

Robert,

I was ... amused by your comments, to say the least, especially because, sad as it is, I consider the possibility that mySQL WILL become DB2 for COBOL z-OS (or some equally absurd configuration) to be disturbingly high.

IBM is a very conservative company that took a remarkably bold step about ten years ago with regard to Linux and open source. This mixture doesn't always bode well (IBM has occasionally been called the graveyard where good companies go when they die, not entirely unfairly).

My suspicion is that the best strategy that IBM could follow with regard to Sun overall is to keep the company (and its branding) as a separate "innovative" center within IBM - rather than trying to parcel things off to one or another of its existing department.

However, given the tight times that all of us are facing right now, I'd be more likely to guess that they'll dismember Sun, even though it's probably not the best move overall. Overall, the loyalty that many people feel to Sun is muted in comparison to their loyalty to a given product (Java, mySQL, OpenOffice, etc.) so the Sun brand really has significance only to those working with OpenSolaris or with Sun servers.

Participate in Sun / IBM Poll:

Would you give money to Sun to keep it independent and developing products?

Vote here:

http://www.misterpoll.com/polls/426985


View current results:

http://www.misterpoll.com/polls/426985/results

Participate in message forum:

http://www.misterpoll.com/forums/192808

Making it easier to use ZFS would also be a boon.

Sun, however, has been reluctant to completely open source Java, creating licenses on the language that still keep the evolution of the language firmly under Sun's control - and as such out of bounds for most Linux distributions to include in a purely GPL/LGPL stack.

Java is already available under GPL 2.0 and has been since November 13 2006.

This article is so full of misleading statements, I just don't know where to start:

1) Java is completely open-source, under the GPL.
2) Java is on several Linux distributions, including (but not limited to) Ubuntu, Fedora, and Gentoo. OpenJDK has even been ported to FreeBSD, and it's being ported to Mac OS X.
3) OpenSolaris is totally open source, with a completely open source distribution starting with OpenSolaris 2008-05. Despite your claims that it's not really open source, FreeBSD and Mac OS X have already integrated its technologies, including ZFS and DTrace.
4) It is possible to run Linux programs on OpenSolaris with Brand X (or is it Brand Z). However, it's cumbersome because it requires the installation of a Linux distro inside of it, and it's very clumsy to set up.
5) IBM will improve Java?!? Right. It took them a year later than Sun to release their JDK 5.
6) IBM will not improve OpenOffice. Look at the mess that is Lotus Symphony.
7) IBM has lots more money that Sun, but somehow, their software sucks much more. Compare Glassfish to Websphere. Compare Sun Java System Messaging Server to Lotus Notes. Compare the userland of Solaris to AIX (where some standard Unix commands like diff have known severe bugs). Heck, even compare Netbeans to Eclipse (granted, IBM doesn't control Eclipse and it's improved dramatically since the community took it over).

As I've said before, IBM will ruin or kill all of Sun's great software products. IBM is excellent at selling other people's software (Linux) and engineering their own software purely so that it meets bullet points in presentations to clients. It's a nightmare to maintain everything from ClearCase to DB2. How IBM manages to do far less with far more than Sun is absolutely beyond me.

Oh, and don't forget that Websphere has been proven to break the JEE spec, according to posters to JavaLobby who deployed a specific webapp on both Websphere and other standard containers.

I can tell you why ibm's products suck...they are overglorafied project managers in and the biggest offshore house. When their strategy is to send everything to india for dev, your products will suffer.

Microchannel
OS/2
Lotus
Project Zero (http://www.javalobby.org/java/forums/t98213.html)

*cough*

And I agree with bart, too.

Sun has done the correct thing in creating an open source version of Java while keeping the Official Version to itself. Java is so awesome because it takes user suggested improvements, analyzes them and throws out the junk. Most Pure Open Source projects either suck in the operation or in the User Interface or both.

I will make a prediction, turn Java full open source and it will die.

I look forward for SWT inclusion in JAVA. Ditch AWT/Swing altogether. Make it an add-on. IBM could have the resources to reshape OpenOffice to use SWT. Maybe create a browser with plugins like Firefox. Then we would have a Runtime, Browser, Office and plugins for both. Browsers are geared towards becomming VM platforms, so why not use the JVM for that :)

This is bad news. My experience with IBM is they are incapable of properly following anyone else's or an open specification. If you think this is a good thing you have never had the pleasure of using IBM's very expensive software, websphere, MQ, AIX, rational, db2, etc. I can outline in many ways how bad this stuff is. It is overly complicated, bug riddled software with poor documentation. this is how they rake in consulting fees, I can tell you the company I work for is their 2nd or 3rd largest customer, I think only the US government spends more with IBM. Websphere 6 is slow, buggy, and cumbersome to use, also be careful of ERROR: 645ER37ENT, it is a distributed app but the errors and configuration look like mainframe. I agree with Wolfmandragon that open source is great but for corporate use there needs to be a different tree and packaging it just works better.

I have a better idea, sun stays independent and all of IBM's customers switch, IBM because of 3 letters well known to stupid know nothing managers charges easily in excess of 3x the money for equivalent products, in terms of hardware we recently started switching to sun servers because they are that much cheaper, sun will sell us a complete T5240 server for what ibm wants to install one power5 chip. This is the same IBM from years ago, not a tech company but a Borg like MS.

I also thing that in the large unix server space this raises big red anti-trust flags. With the completion of this merger IBM wields control over ~70% of the 12bln per year unix server sector. I really think this merger is about getting control over java, killing a big threat to its websphere franchise(glassfish), and most important right now IBM wants a piece of sun's unix server business as recently they have been beaten by sun on new contracts with IBM customers.

For the benefit of Open Source and java I think it is imperative Sun remains independent.

Eclipse is the only piece of IBM software I know that works easily. Their Rational products can't even be deployed on a larger scale. Their web site is a mess. Their licensing is as opaque as Microsoft's. Nobody wants IBM's JDK, but you are forced to use it for some of their products. WebSphere is a monster I wouldn't like to take on. And all of their stuff is way too expensive. There's no dual licensing so you can try stuff. Their support doesn't even get back to you. IBM? No thanks!

Java is great _because_ of Suns leadership. IBM would ruin it.

Some other questions : What will happen to

-Glassfish vs Websphere
-Netbeans vs Eclipse
-Swing vs SWT
-JavaFX, which is a layer above Swing, and is promising but buggy

News Topics

Recommended for You

Got a Question?