When considering how best to address the rumors – at that stage, anyhow – that Oracle was in talks to acquire JBoss, Sleepycat, and Zend, I debated whether or not to split them off into separate entries. Not simply for the sake of brevity , but also because they seemed to be driven by very different factors. The more I thought about it, however, the more I convinced myself that they belonged together, that to discuss them independent of one another was to miss the contextual glue that binds them together. So without further delay, here’s the Q&A:
Q: Ok, first, what’s the current state of affairs? Who’s actually been acquired, and who’s simply rumored?
A: Well, as previously discussed, I can only comment on what I actually know – what I either know is public or what is more or less open, acknowledged rumor. I’m not a reporter, and I’m not looking to break news, so as I’ve said before – you’ll probably hear it here last. That said, here’s the situation that I’m aware of: Sleepycat has been acquired, most people here at OSBC seem to believe that JBoss will be (and the late breaking rumor is that the deal is done – thx, Cote), and in one panel yesterday Zend’s Doron Gerstel did little to deny the Oracle interest. But so far, only Sleepycat is a definite, at least as far as I’m aware.
Q: How likely is it that the other two close?
A: No idea. The reporters would have a better read on that.I just analyze the news; I don’t break it.
Q: Before we get into the specifics of the individual transactions, let’s deal with the general. Why acquire all three of these players?
A: While each acquisition is (in the case of Sleepycat) or would be (in the case of JBoss and Zend) be driven by a different thesis, to me there are a couple of themes that tie the seemingly unrelated moves together.
- Open Source:
Most obviously, the moves – either individually, or in concert – would be perceived as a validation of the reality of open source within the market place. I say perceived, because there’s one school of thought that says at least one of the moves – Sleepycat – is actually an attempt to undermine an important open source competitor. There is considerable skepticism, then, as to whether Oracle – who candidly does not enjoy a sterling reputation in its dealing with some open source players – would be simply embracing open source, or embracing as a mere prelude to extending and extinguishing. Irrespective of how things played out with the various communities they’d be buying their way into medium to long term, however, Oracle would be acquiring some immediate open source credibility and skills. That could be leveraged.
- Volume Opportunity:
It would be a recognition on the part of Oracle that future growth will not solely be through the traditional ‘enterprise’ sale, where ‘enterprise’ means high margin, high cost, high complexity engagements, but through lower cost, lower margin volume opportunities. Opportunities such as those targeted by vendors such as JBoss, Sleepycat, and Zend. Oracle apparently sees this – and it likely played a role in their reported attempt to purchase MySQL.
- The Best Defense is a Good Offense:
The interesting thing about the acquistions, with the possible exception of Sleepycat – although I could make the case there too as well – is that besides opening new doors to Oracle’s business, they’d have destablizing effects on its competitors businesses. The Peoplesoft and Siebel acquisitions are good examples here; besides giving Oracle immediate and major presence in the packaged applications market, they simultaneously weakened – at least in terms of perception – competitors such as IBM and SAP. These acqusitions are similar, as we’ll explore later.
- Developer/Community Relations:
I don’t know if Oracle would cite this as an acquisition driver, but I think this one is potentially big. If you’ve read this space previously, you’re aware that I’m quite convinced that we’re in the midst of a fundamental shift in how enterprises buy technology. It’s not that the days of the high priced software salesmen are gone, though I’ve heard that argument made more than a few times in the past couple of weeks, but that the importance of developers is finally being recognized. Where the CIO was once the product kingmaker, these days its the grunts in the field. Witness the success of Linux, MySQL, etc. Oracle, like many of its enterprise software brethren, has not enjoyed the best relationship with developers given the traditionally high price and complexity of its products – the barriers to entry. There’s a certain amount of cultural disconnect as well. All three of the firms being discussed enjoy solid if unspectacular reputations  with the developers that I speak with, potentially giving Oracle an entrypoint to a variety of communities of highly important technologists, communities that are currently closed to the vendor.
- Product Synergies:
While one doesn’t typically think of JBoss and PHP as functioning side by side, the fact is that they’re both increasingly a part of the heterogeneous network environments we see – and in fact I’ve spoken to two folks recently that are using both. So that then begs the question: does having them under one roof offer opportunities for closer ties and better integration than could be forged via mere partnerships? My answer is maybe. I’m not persuaded, but neither do I think it will hurt. Alfresco’s Matt Asay goes further and speculates that Oracle could enter the open source stack integration business, thus competing with the likes of OpenLogic, SourceLabs and Spikesource.
There are more angles that we could discuss, but those, as far as I’m concerned, are the most crucial ones.
Q: Onto the specifics. Let’s take look at the individual transactions, beginning with the only one consummated to date – Sleepycat. What’s the rationale behind that move?
A: The popular answer thus far has been that this is a move aimed primarily at weakening a competitor – MySQL – rather than strengthening Oracle. Assuming that the aforementioned attempt to purchase MySQL occured more or less as reported, it seems safe to conclude that despite claims to the contrary, Oracle does in fact view MySQL as a competitve threat, one of some significance. Not at the high end, but a threat nonetheless. Although Mickos declined to specify when Oracle attempted to buy them out, my bet would be that they were Plan A. Plan B, according to many, is to cut off MySQL’s oxygen supply by acquiring the components it relies on – first Innobase, and now Sleepycat.
db4objects’ Christof Wittig, himself the head of an open source database firm, subscribes to this notion, saying:
Why would Oracle buy Sleepycat. They are not really in the embedded space. Well, they are, but they are laggards with 2% market share. Buying another 2% doesn’t really make sense.
Oh, hang on. Isn’t Berkeley DB inside MySQL? Aren’t we repeating the InnoBase move – and this time, hit MySQL really hard?
In speaking with various parties at OSBC, this is more or less the conventional wisdom, and I can see it. Would Oracle love to weaken MySQL? Does the pope wear a big hat? But the really interesting question to me is: does this accomplish that? I have my doubts. Just as I said when they acquired InnoDB, I think the long term implications for MySQL are less than dire.
Q: Why is that?
A: Well, I think the doom and gloom predictions of MySQL’s future following these acquistions are somewhat premature. Yes, it’s true that to some degree MySQL is reliant on both InnoDB and Berkley DB, and yes it’s true that license negotiations with both parties are likely to be a lot more interesting the next time around, but while the ground under them is a bit softer, MySQL’s still dealing from a position of strength. Their community is still immense, and it’s clear that they know it. At a (too large) panel at OSBC on Tuesday night, Mickos responded to questions about the Oracle issue by saying that hitting MySQL in this fashion was like “trying to kill a dolphin by drinking the ocean.” Quotable, as usual. But the point here is that MySQL’s ubiquity gives it breathing room – a margin for error. If, 12 or 18 months from now, Oracle still owns key pieces of their infrastructure, I’d be more concerned, but for the time being I think they’re fine. In the meantime, I’d expect Marten, Zack and co to be exploring options internal (throwing engineers at MyISAM) to external (Progress or Solid). In any event, I think arguing that this is purely an anti-competitive move on Oracle’s part is illogical given the quality of some of the Sleepycat assets.
Q: How so – aren’t they just another embedded database vendor?
A: While they’re not as high profile as a platform like MySQL – possibly because it’s not the traditional SQL based RDBMS – Berkeley DB’s a very interesting technology. As Sleepycat’s Rex Wang says in a comment to Mike Olson’s announcement post here:
We conservatively estimate that there are 200 million copies of Berkeley DB running out there. Berkeley DB is in every copy of Linux, every distro, whether it’s server, desktop or embedded. It’s also in every copy of BSD Unix, so OpenBSD, FreeBSD, NetBSD, etc. Every copy of OpenLDAP and sendmail uses us. Every copy of OpenOffice and StarOffice (Sun estimates 40 million are out there). 3 million Motorola cellphones. Many copies of Movable Type, Subversion and Apache HTTP server.
And he leaves out some of the very large, big name clients they’ve won over the years. Were they driving monstrous revenues? Nope. But it’s not as if it’s a seldom used, throwaway technology, either. It’s also interesting to consider what Oracle might do with their XML store bits – the opportunity to be the next MySQL, after all, is out there.
Q: That’s probably enough on Sleepycat – let’s move on to the uncertain parts of the deal. Why might JBoss make sense?
A: Before acquiring the likes of Peoplesoft and Siebel, what did Oracle do for a living? Sell databases – lots of them. Not cheap databases, not simple databases, but highly functional, highly scalable RDBMSs for serious usage. Now ask yourself what’s the most common development language/platform used in those scenarios? Java. Next, ask yourself what’s closest equivalent to MySQL in the middleware space in terms of ubiquity. It certainly isn’t Oracle’s application server – even Larry Ellison admits that, saying “We are clearly not No. 1 in middleware.” Java deployment sells databases, JBoss is the choice for a great many Java deployments, and the economics work – that alone seems to be a decent basis for acquisition.
Q: What do you mean the economics work?
A: For any given project, there’s a software cost, a hardware cost, and a services cost. Over the past few years, there’s been a lot of downward price pressure on all three due to a variety of factors, but the software cost in particular. So if Oracle wants to continue to make good money on the database component of that overall software cost, it’s in their best interests to ensure that the other components of the software sale cost less. Pitching JBoss as the middleware is one way to accomplish that.
Q: Does Oracle really need another middleware stack to integrate? Won’t the addition of JBoss create more complexity when their Fusion middleware strategy is already enough of a hairball?
A: Only if they let it. If I’m Oracle, I acquire JBoss and don’t let the Fusion guys anywhere near it. I let them do their thing, giving the small firm access to big company resouces in its effort to compete more effectively with the Genonimo/WebSphere CE combination. Are there messaging and sales issues in maintaining – effectively – two competing middleware stacks? Sure, but I think the benefits outweigh the risks there.
Q: Speaking of IBM, what’s the impact there? Would JBoss be an attempt gain share at their expense?
A: Absolutely. I’ve contended, for example, as have many others, that the IBM acquisition of Gluecode and the resulting development efforts being poured into Geronimo were more or less a direct response to the success of JBoss. In my Q&A on that acquisition, I wrote that:
Longer term, JBoss is facing a new sales challenge; they knew how to sell against webSphere, and know how to sell against Gluecode as it exists now (see here for Marc Fleury’s take on the news). But everyone remembers when WebSphere was merely a speck in WebLogic’s rearview mirror.
The point was that in my view, JBoss would have a difficult time competing against IBM’s greater resouces over the longer term – that’s one problem that Oracle could solve. There’s no question that giving the popular JBoss stack greater resources to draw on is not a prospect likely to excite any of the WebSphere folks – Community Edition or otherwise.
Q: Last company to discuss – what’s the deal with Zend?
A: The same logic used to justify the JBoss acquisition applies here, maybe even to a greater degree. In other words, the logic is simple: a.) there’s a lot of PHP in the enterprise, and may be more, b.) PHP apps usually require some persistence mechanism, i.e. a database, and therefore c.) owning the firm mostly closely associated with PHP would provide growth opportunities at least for the database arena – and potentially the biz apps market longer term. While PHP is not traditionally considered middleware in the sense that, say, JBoss is – the fact is that LAMP(HP) is increasingly viewed as an alternate stack to build from. Middleware, in other words, albeit of a different stripe than JEE.
Q: How about Zend as a lever against competitors – are there opportunities to exploit there as well?
A: Potentially, although in a less direct fashion then say the pickup of JBoss might allow. IBM, for example, was quicker out of the gate in seeing the opportunity around PHP and partnering with Zend than was Oracle – but might Oracle pull the rug out by taking the next step? In terms of perception, the answer I think is maybe. PHP is and will remain open source, so it’s not as if Oracle could lock IBM out even if were to choose to try. But could it brand itself more closely with the popular dynamic language than could, say, IBM? Or Sun? Yup, because having folks like Andi Gutmans on staff allows for that.
Q: I’m glad you mentioned Andi, because that begs the question: what importance do you ascribe to some of the developers involved in the various firms?
A: Substantial importance. I’ve been asked multiple times in connection with these rumors whether or not it’s possible to ‘buy’ a community, and my general answer is no.
Q: So you don’t believe that Oracle can, in effect, buy the respective JBoss (and associated communities where it employs devs), PHP and Berkeley DB communities?
A: No, I don’t. It’s not really a question of licensing (JBoss is LGPL, PHP is BSD, and Berkeley DB uses the OSI approved Sleepycat Public License) – or forking (Sleepycat, for example, owns their IP which why they are able to dual license their product), it’s more a question of community dynamics. The attractiveness of the above projects is predicated not solely on their technical merits – maybe not even primarily – but on their popularity. If any vendor thinks that it can uniformly impose its will on a large scale community of developers, they’re in for a surprise. Those developers, in most cases, have alternatives they can and will turn to – Geronimo in the case of JBoss, Ruby/Python/Perl in the case of PHP, and a variety of datastores with respect to Berkeley DB – should they begin to get the impression that they’re being pushed around.
Q: What’s the point of acquiring them, then?
A: Well, apart from the technology, there’s the people as discussed earlier. There are numerous examples from the past few years where non open source companies have sought entry into the open source market not by purchasing communities, but by purchasing influencers in those communities. As I told Martin last Saturday, “you can buy influencers (such as lead developers) in an open-source community. That’s a viable strategy.” One of the more notable examples of this in recent years is Novell; which transitioned itself virtually overnight by acquiring SuSE and Ximian; Miguel and Nat alone make Novell relevant in conversations where they previously would never have been discussed. So while commercial entities might not be able to buy and direct communities as they might wish to, they can certainly obtain influence on the same communities – they just need to be careful how and when they use it.
Q: What could go wrong in any/all of the aforementioned transactions?
A: M&A is never as straightforward as it seems on paper, and is especially challenging when the cultures of the merging organizations are not closely aligned. I remember reading an interview with Cisco’s John Chambers years ago in Business 2.0, I think, in which he described cultural meshing as the #1 consideration in acquisitions that represented anything more than just a straight grab of technology. As Matt says:
This is easier said than done, of course. I learned at Novell that the shift to an open source-friendly culture and business model is non-trivial. It will not come easily to Oracle (or any traditional software company), $7B in incentive dollars or not.
There’s a complicating factor here as well, in that besides the simple cultural delta between Oracle and the organizations in the crosshairs, there’s the larger challenge of making the acquirer more open source friendly. The commenters on the Sleepycat blog, for example, expressed some sadness at the acquisition and I’ve heard from a couple horrified PHP developers, who are concerned that Oracle’s culture will bleed into their platform of choice. So the challenge is not only to assimilate the new organizations, but recombine some of their DNA with Oracle’s own. As Matt says, that’s easier said than done. Sun’s Bill Vass said as much on a panel yesterday, in that explaining to your salespeople why their high margin product being made free and open source is a good thing is not a trivial exercise. And if there are any more aggressive salespeople on the planet than Oracle’s, I haven’t heard of them.
Q: What do you think of the moves, either separately or individually?
A: I think they’re bold moves, and would have at least the potential for transforming Oracle into a more open source friendly organization. It’s not without a boatload of risks, but it’d be interesting.
Q: Anything else to add?
A: Just that I hope the friendliness and openness of some of folks already acquired or rumored to be rubs off on Oracle, because while I’m not sure how they are with other analyst firms, candidly they have not always been easy for us to interface with.
Disclaimer: This is a bit more complicated than usual with the M&A and uncertainty, but let’s see: db4objects, IBM, the newly acquired Sleepycat, Sun and Zend are RedMonk clients, while Alfresco, JBoss, MySQL, Novell and Oracle (with the exception of Sleepycat) are not.
 It’d be a bit much to expect anyone to buy the argument that I’m concerned with being concise 🙂
 I’m aware of the frustrations with JBoss around the trademark issue, but you can’t argue with its continuing popularity.