Of course Sun got acquired yesterday: it was a holiday for me. Just as the skies are sure to open up when I’m about to get on a plane, so too are acquisitions inevitably made when I’m not supposed to be working. True, the acquirer was not IBM as many had predicted, but rather Big Blue’s database nemesis, Oracle, but that’s how these things go. It’s never over until it’s over. And sometimes – perhaps even now? – it’s not over when it’s over.
Assuming that it is, in fact, over for Sun Microsystems as an independent entity, the question is: what’s next? With so many moving parts, the deal is complex in any event, and all the more so here because it could have a transformative impact on the enterprise software, storage and server markets, not to mention several large and highly visible open source communities.
To explore the ramifications of the proposed deal, let’s indulge in the traditional Q&A.
Q: Before we begin, is there anything to disclose?
A: Yes indeed. Sun Microsystems, at present still an independent entity, is a RedMonk customer, as were acquisitions such as MySQL. Other mentioned parties, including Canonical, IBM, Microsoft, and Red Hat are RedMonk customers as well. Oracle, meanwhile, is not. That should cover it.
Q: For those that may have been living under a rock, can you summarize the news?
A: Sure. Yesterday, in an effort to screw up my Patriot’s Day plans, Oracle announced its intention to purchase Sun for $9.50 a share, putting the valuation at $7.4B (or $5.6B less cash on hand). Subject to the customary closing conditions, it would appear that Sun’s days as an independent will be over by this summer.
Q: Was this a surprise?
A: Not particularly. Rumors have been swirling for weeks about the fate of Sun, following its highly visible on again / off again talks with IBM. While Oracle wasn’t the most recent rumored acquirer, there’s a long history of Oracle-as-potential-acquirer speculation. There is little question that the public knowledge of the IBM/Sun talks damaged the CA based systems vendor. Sun was facing questions about its long term viability from customers even prior to the acquisition talks due to its recent financial misfortunes, and in the wake of the breakup of the IBM talks several customers I spoke with had indefinitely shelved and/or cancelled their conversations with Sun salespeople due to the uncertainty.
In this context, an acquisition may well have been inevitable, and while Oracle is far from the perfect fit, it’s a) got the financial resources for such a transaction and b) represents a logical home for a portion of Sun’s assets.
Q: Could Sun have continued to play as an independent?
A: For the short to mid term? Sure. They had almost two billion in the bank, and had taken their headcount down significantly in the first quarter, so they were not likely to fail in the next quarter or four. But Sun’s attempts at staging a comeback may well have been done in early this fall when the economy wrapped itself around a tree and its traditional high end customers put traditional high end purchases on hold, because as much as Sun’s ambitions were to sell to a volume audience, its execution in that respect never matched its aspirations and it was therefore disproportionately punished by the high end slowdown.
Q: What do you mean by its ambitions to sell to a volume audience?
A: While Sun’s initial success and fame was the product of its workstation business, its peak was largely fueled by its status as the enterprise server/operating system platform of choice. In the years leading up to the dot com implosion, Sun’s Solaris operating system and SPARC servers were the standard by which all others were judged. In the years since, however, a combination of competition – x86 Linux servers, primarily – and Innovator’s Dilemma-style missteps with open source, Solaris, and SPARC saw that marketshare dominance evaporate.
As a result, Sun’s days as a high-flyer were a memory, and they were back to playing the pesky upstart, but one saddled with big company obligations in headcount and real estate. Over the past few years, as Sun’s leadership transitioned from the change averse McNealy to the one time head of Sun software, Jonathan Schwartz, we’ve seen change. Not as aggressive change as we’d like, it must be said, but many of the missteps – not open sourcing Solaris, trying to fight the x86 tide, and so on – have been corrected.
But as much as the product catalog had been overhauled to appeal to audiences outside of Sun’s typical high end telco or financial service house, its sales and procurement process remained tragically antiquated. The consistent complaint we’ve heard from potential smaller customers (including RedMonk) for years has been the same: we can’t buy from Sun. Even today, in spite of well intentioned programs like Startup Essentials, the end result is the same: it remains difficult for would-be customers to engage with Sun. A friend who works for the government had difficulty recently procuring a quote for several hundred thousand dollars of storage equipment: that’s a problem.
Though it was Jonathan’s stated ambition to sell to customers that no Sun rep had ever spoken with, then, the firm could never quite get there. Without those kinds of volume, low margin sales – the kind that have kept Dell afloat, Sun was left to the mercies of its high end buyers, with the end result we saw yesterday.
Q: Was Sun’s issue really volume sales? Or a diversity of product focus that wasn’t sustainable?
A: Sun’s biggest issue, as mentioned, was the deprecession. But independent of that, certainly focus was a problem. We’ve recommended in the past that Sun decommit from underperforming assets that have either zero or a negative return, and it should be expected that Oracle will be ruthless in this regard, both because they lack the emotional attachments to the assets in question but also because they are not required to invest in certain products for visibility purposes: Oracle is a household name with the returns to match.
Q: Setting aside the past, and how Sun got here, what does the future hold for the assets of the firm? Are they worth the purchase price?
A: Oh, they’re undoubtedly worth it. I joked when discussing the potential for IBM to aquire Sun that I’d contemplated purchasing the firm. When it was trading near its cash position, after all, anyone capable of securing the loan of a few billion was likely to profit even if your intention was merely to strip it and sell it for parts. Oracle obviously paid a substantial premium over the cash on hand, but the value of Sun’s assets remains substantial, in my view.
Larry Augustin’s view, in fact, is that Oracle will spin the hardware assets – storage and servers alike – at a sufficient cost to make the acquisition of the Java, MySQL, Solaris etc assets effectively free. While I don’t necessarily agree with that, it serves to illustrate the point: Oracle stands a decent chance of profiting on the acquisition. Mike Dolan, for his part, thinks that Oracle will meet its profitability targets simply with further workforce reductions.
The acquisition conversations alone, in fact, demonstrate the value of Sun’s assets. Neither IBM nor Oracle are unintelligent or careless in their acquisition process; both clearly saw value and unrealized potential in what Sun had built.
Q: But value aside, what’s to become of the assets? Let’s start with the most obvious in Java?
A: It’s difficult to say. Not the future of the language and platform, of course: Oracle, like IBM, is heavily invested in a Java world, from the tools back up to the middleware portfolio. And clearly Oracle understands the importance of the ecosystem and is unlikely to rock the boat via draconian or dictatorial changes to the JCP. That said, it remains to be seen whether or not large Java players such as IBM will be as comfortable being strategically beholden to a clear and present danger like Oracle versus the smaller and therefore less threatening Sun.
Could we see, for example, an IBM developer strategy that while still Java focused, incorporates rather than attacks the increasing diversity of deployment models? This acquisition is likely to have them at least asking that question.
However the politics resolve themselves or not, though, the Java landscape is characterized as much by its inertia as anything else. Ergo any potential shifts in the landscape here are likely to be glacial in pace. I’d expect something very akin to the status quo for the foreseeable future.
It’s worth mentioning, as well, that one anonymous commenter believes that the embedded business will be “the surprise benefit of the acquisition to people who think of Oracle as only an enterprise software company.” Given some of the embedded Java revenues, I don’t know that the benefit would surprise me all that much.
Q: Could Oracle possibly improve on Sun’s stewardship of Java? Might they, for example, address the Field of Use restrictions that limit the Harmony project as Ryan speculates?
A: It’s certainly possible that Oracle, with less emotional attachment to the Java platform, could address some of the common complaints levied against the JCP. But I am considerably less optimistic that Oracle will be creative enough to see the opportunity lost to the FOU issues, rooted as those are in fears over embeddable business revenues. Oracle’s focus will always be on what’s best for Oracle first, and while the firm has tactically embraced open source, it has not traditionally done so strategically. We have yet to see an Eclipse moment from Oracle, remember.
Q: Ok, how about Solaris? What does the future hold for that operating system given the fact that Oracle already sells a competing offering (Oracle Linux, the Red Hat clone)?
A: This is a fascinating question, to me. While Solaris was mentioned prominently on the call discussing the acquisition, I found it notable that the FAQ they distributed is, at best, ambiguous:
The Sun Solaris operating system is the leading platform for the Oracle database. With the acquisition of Sun, Oracle can optimize the Oracle database for some of the unique, high-end features of Solaris. Oracle is as committed as ever to Linux and other open platforms and will continue to support and enhance our strong industry partnerships.
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.
Translation: we’re not telling you what we’re going to do. Which is to be expected, of course.
The betting here is that Solaris will continue to be supported, but not as a frontline option, with the possible exception of cloud offerings where the quirks of the operating system are rendered invisible by the platform. Think IBM with AIX, HP with HP-UX, and so on: there is ample precedent for the (successful) continuation of two competing product lines, and as Oracle itself acknowledges above, there’s an awful lot of Oracle running on top of Solaris.
The real question here, however, is whether or not Oracle will consider the relicensing of Solaris. At present, remember, the open source version of Solaris is licensed under the CDDL, which is incompatible with the GPL and thus with Linux. As I’ve said before, I believe the choice made sense at the time:
Suffice it to say that on the one side we have the contention that the license was selected purely for its merits, and on the other we have the assertion that it was chosen at least in part for its incompatibility with the GPLv2. Regardless of which side is correct, I’m of the opinion that the license chosen at the time was a necessary choice to ensure its survival. With no one convinced at the time that OpenSolaris would survive as an open source project, I think the GPLv2 compatibility would have allowed Linux to cherry-pick from technical differentiating features like DTrace and leave the project for dead.
But while it may have made sense at the time, it’s an open question whether or not it makes sense indefinitely. What Oracle thinks is an even bigger question.
Q: How do you mean?
A: Just as it does with MySQL, Sun owns the entirety – with the exception of a few encumbrances – of the copyright to the Solaris codebase, as anything it didn’t write itself was licensed per the terms of the Joint Copyright Agreement (JCA). In practical terms, this means that Sun – and by extension, an acquirer – would have the right to relicense the asset.
Q: What from Solaris would Oracle be interested in relicensing?
A: That’s an excellent question. DTrace, given the ongoing struggles with SystemTap, would be one very interesting piece of technology (and it’s already been ported, sort of). And one wonders about ZFS as well. What I wrote in the past about IBM potentially owning the rights to ZFS is just as applicable to Oracle:
Take something as mundane as the filesystem, ZFS. While it’s not perfect – I regularly field complaints about its file creation, for example – it is nonetheless the design target for the next generation Linux filesystems, which aim to meet or exceed both its performance and its functional capabilities (think snapshotting, pools, etc). The primary problem is that even the bridge filesystem, ext4, is not yet production ready, and the ultimate response, Btrfs, is probably not going to be ready for testing for a year, meaning it’s years, pluarl, from seeing production workloads. Owning the copyright to ZFS, however, would allow IBM – if it so chose – to relicense the asset under the GPL so as to be compatible with Linux, thus permitting a port.
So now that it would appear that Oracle will have at least the legal right to relicense these assets such that they would be legally portable to Linux, there are many questions that can’t yet be answered: 1.) would Oracle, in fact, relicense the code? 2.) if they did, what impact would that have on Oracle’s Solaris business? 3.) would Linux accept the additions, given the occasional NIH symptoms that community displays (just as the Solaris/OpenSolaris community does, to be fair), 4.) if they’re accepted, what impact would this have on Oracle’s commitments to BTRFS? 5.) what impact would it have on SystemTap? And so on.
It’s too early to say on any of the above, but it can’t be debated that the acquisition of Sun by Oracle would spell change for Solaris, and – potentially – it’s end as a mainstream operating system offering.
Q: Moving on, how about the impact to MySQL? Is this the death of that product?
A: I don’t believe so, no, nor does ex-MySQL CEO Mårten Mickos who I spoke with yesterday. Mårten sees this, at least in part, as a response to the Microsoft competitive threat, as he told Forbes. It seems extremely unlikely to me that Oracle – who’d been a rumored acquirer of MySQL before Sun snapped them up – would acquire the business just to kill it, particularly since that MySQL’s phenomenal growth has not come at the expense of their own relational database product. They’ve already played in the MySQL acquisition space, remember, buying InnoDB back in 2005 and maintaining it effectively.
In the abstract, it’s easy to see how Oracle could use MySQL as a complement to its flagship product; MySQL for low end, low margin accounts where Oracle doesn’t currently compete or, tactically, to undermine SQL Server and/or DB2 in particular accounts. The products are the perfects foils for one another, in many respects: MySQL is to startups/developers as Oracle is to the enterprise mission critical workloads.
Practically speaking, however, one can see obvious problems ahead. As a few folks have pointed out, the Oracle sales force is going to dislike MySQL even more than the Sun sales force did, as it is a low margin product competing – at least in some sense – with a high margin staple. The other question is the reporting structure: will Oracle fold MySQL into their database division? Or keep it reasonably autonomous? The former would likely be disastrous, the latter is difficult to imagine given a.) the political landscape and the absence of strong leadership at MySQL. While Karen Tegan Padir is well thought of within MySQL and the hand chosen replacement for Mårten, I agree with those who predict that it would be difficult for her to compete successfully with the entrenched Oracle database chieftains.
Q: Is this a boon for other database projects, then? PostgreSQL? Drizzle?
A: Probably not. MySQL is a phenonenon unto itself at this point, and Oracle would have to work hard to derail its momentum in the near term. Will we see some defections to PostgreSQL? Perhaps; I’ve already seen developers discussing the possibility. But by and large, database migrations are difficult, particularly in a world where everyone knows MySQL and everything works with it.
As for Drizzle, while I remain a believer in and follower of that project, it’s not ready for prime time just yet, and in any event is a different project with different desogn assumptions.
More interesting to me will be whether the more mainstream forks of MySQL gain traction. Percona, OurDelta and so on could benefit strongly from this acquisition, though as Mårten pointed out yesterday, Oracle will maintain ownership of the MySQL trademark which inhibits Percona et al’s ability to effectively market their offering. As long as Oracle adheres to the dual license model favored by MySQL and subsequently Sun, they would be operating at a technical disadvantage to forks that can incorporate all available assets. Here’s how I’ve discussed this in the past:
Generally, [the dual license] model has served MySQL fairly well. By controlling the intellectual property, they retain the rights to relicense the code, thus protecting a revenue stream. They also were afforded a slightly greater protection from forks versus more collaboratively developed projects like Linux, in that they – theoretically – employed the majority of the people qualified and paid to work on the codebase at the lowest levels. But let’s come back to that.
What’s the catch to the model? In part, it’s that the burden of development is born almost entirely by the MySQL staff, but the more relevant concern here is the inability to consume external contributions – even if they’re excellent – without licensing them.
Stated more simply: as long as MySQL remains committed to the dual licensing model, it will be unable to accept the same patch set that open source only versions of the code can, because they do not share the same licensing concerns. Which is why we’ve seen these spring up, and possibly why the MySQL-derived Drizzle project has taken a different approach from its parent.
To be clear: we’re a long way, in my opinion, from Percona and the other MySQL distributions gaining mainstream traction. But it’s not impossible to envision a scenario in which one gains mainstream acceptance because of its technical superiority and/or support and acceptance from a large, Oracle competitor.
Q: Ok, let’s head to the lightning round. Predictions for other Sun assets: Glassfish?
A: Could be Oracle’s equivalent of WebSphere CE, i.e. a lower end alternative to WebLogic that Oracle could use to stave off incursions from the JBoss threat.
Q: The Identity business?
A: Oracle has its own identity line, so it could be purely Darwinian. But given that Sun’s product here is well regarded and profitable, it seems likely that it will survive the acquisition.
A: While there’s little love lost between Oracle and Microsoft, the database giant is considerably less Quixotic than their newly acquired friends, so I would be surprised to see the OpenOffice.org asset maintained. I’d look for it to be either sold (to IBM?), spun into a foundation (depending on the cost), or left to sink or swim on its own.
A: That’s a tougher call. Oracle is an Eclipse shop, but their JDeveloper product is not Eclipse based, as Mike Milinkovich pointed out on Cote’s blog. NetBeans is the superior technology – at least when dynamic languages are factored in – but will Oracle be as enthusiastic about maintaining a second sizable open source tooling project when they’re already strategically invested in another? It seems unlikely.
Q: xVM and VirtualBox?
A: An intriguing question, given the fact that Oracle’s virtualization capabilities are primarily limited to their rebadged RHEL, which is shifting to KVM – a competitor to the Xen project the xVM is based on. While virtualization is clearly a critical technology, it’s not clear that Oracle would view this as a profitable market. So, the future here is uncertain.
Q: What should we expect with respect to layoffs?
A: Sadly, because I know and respect many of the those who work at Sun, I think the layoffs will be swift and severe given the profitability Oracle has promised the street. Though Sun employees have suffered through several rounds of layoffs in the last few years, Oracle will be entirely unemotional in their evaluation of assets both people and not, and there will be a great many people looking for jobs.
Q: What about the question, raised above, of whether or not Oracle intends to spin off the hardware assets?
A: I think the cloud, honestly, is the wildcard there. While Oracle has started telling its own cloud story, Larry Ellison has been a noted skeptic of the trend:
The interesting thing about cloud computing is that we’ve redefined cloud computing to include everything that we already do. I can’t think of anything that isn’t cloud computing with all of these announcements. The computer industry is the only industry that is more fashion-driven than women’s fashion. Maybe I’m an idiot, but I have no idea what anyone is talking about. What is it? It’s complete gibberish. It’s insane. When is this idiocy going to stop?
If he remains committed to that position, unlike HP, IBM, Microsoft and the newly acquired Sun, it makes little sense – in my view – for them to own the hardware. Yes, there was a great deal of messaging at acquisition time of the ability for Oracle to sell customers stacks “from application to disk,” but I find it difficult to believe that Oracle would maintain the hardware businesses solely for the purposes of getting into the appliance market.
If, on the other hand, Oracle foresees a future in which its larger customers will begin migrating strategic portions of their workload to HP, IBM, and Microsoft, it might behoove it to have their own platform. One built on Sun hardware and Sun cloud technologies, perhaps. Given that the cloud collapses enterprise buyers choices – hardware, operation system, database and, in future, application – there could be no more critical question following the acquisition than this. Thus it will be interesting to see which way they head.
Q: What about the argument that Oracle is a software company, and doesn’t have the DNA to run the hardware side?
A: There’s no doubt that, as Cote argues, being a true systems company is a pretty trick to pull off: that’s why there are so few of them around, and why even a Cisco seems reluctant to fully commit. That said, I think would be platform providers that do not have their own cloud offerings will be at a distinct disadvantage. And while Oracle is currently offering some of their applications up via the cloud, that’s a very different model than selling Oracle Linux, or Oracle 10g, etc via the network.
In other words, if I were in Oracle’s shoes, I would not divest myself of the hardware and cloud offerings, but seek to leverage them as a new and potentially critical route to market.
Q: Is this good or bad for the industry?
A: Well, it’s good for Oracle, so by extension it’s not positive news for competitors such as IBM, Microsoft and Red Hat. Beyond that, the implications for the various open source communities are murky. While strategic projects such as MySQL are likely to survive and remain viable, Oracle does not share Sun’s public commitment to open source and therefore cannot be expected to sustain its commitments in as many areas. From NetBeans to OO.o, it’s possible – likely, even – that we’ll see a dramatic decrease in investments in open source development from what was Sun.
Depending on how you view the importance of open source, that might be bad news indeed.
Q: Anything else to add?
A: That’s all I can think of off the top of my head. Feel free to ask the questions that I forgot to; I’ll answer whatever I can.