By now, most of you are probably aware that Oracle has made the rather momentous decision to enter the operating system market via Red Hat bits. Not by partnering with Red Hat. Instead, Oracle has decided to take advantage of the fact that Red Hat’s bits are ultimately free to obtain and distribute – less the trademarks, of course – to make its own entrance into the market. In effect, it’s entering the operating system market by proxy.
To look at the how, the why, and the impact let’s do a Q&A.
Q: To begin with, any disclaimers to get out of the way?
A: Oracle’s Embedded division is a RedMonk client, by way of the acquisition of customer Sleepycat Software. A variety of Oracle competitors, including EnterpriseDB, IBM, MySQL, Sun and so on are RedMonk customers.
Q: This news has been rumored for several weeks, but things had been quiet until yesterday – are you surprised by the news?
A: Not terribly. About 10 days ago when things started perking up, I was less convinced that something was going on. But as mentioned yesterday, the uptick in the volume of chatter has indicated that something was probably going to happen, although if you’d forced me to guess last week I probably would have said that an Ubuntu parternship was more likely.
In retrospect, however, none of us should be that surprised given that Ellison essentially laid out his rationale for this move back in April to the Financial Times. The interview, which is facinating in hindsight, probed Ellison on a series of questions relating to open source and Software as a Service (SaaS). A couple of the choice quotes:
FT: Is open source going to be disruptive to Oracle?
LE: No. If an open source product gets good enough, we’ll simply take it. Take [the web server software] Apache: once Apache got better than our own web server, we threw it away and took Apache. So the great thing about open source is nobody owns it – a company like Oracle is free to take it for nothing, include it in our products and charge for support, and that’s what we’ll do. So it is not disruptive at all – you have to find places to add value. Once open source gets good enough, competing with it would be insane. Keep in mind it’s not that good in most places yet. We’re a big supporter of Linux. At some point we may embed Linux in all of our products and provide support.
Just like software-as-a-service, we have to be good at it. We don’t have to fight open source, we have to exploit open source. At some point we could very well choose to have Linux as part of the Oracle database server. We certify it, we test it. We could have JBoss as part of our middleware. It costs us nothing. We can do that, IBM can do that, HP can do that – anyone with a large support organisation is free to take that intellectual property and embed it in their own products.
I’ve had this discussion with the CEOs of open source companies. We’ve looked at buying some, some with very high price tags – but since we already have access to all the intellectual property, why wouldn’t we just embed this technology in our technology and provide support.
FT: What are the arguments against Oracle distributing its own Linux version?
LE: They’re not very strong – now that Red Hat has bought JBoss and competes with us in middleware, we have to relook at the relationship – so does IBM. If Oracle were to have its own Linux distribution, or just provide paid support for Red Hat, that’s one thing – if Oracle and IBM both did it, it’s a whole new world. I don’t think Oracle and IBM want to create a second Microsoft in Red Hat. But you can’t – because Red Hat doesn’t own anything, they own nothing. They couldn’t [become the next Microsoft], they own nothing.
Anyhow, to get back to the original question, no, I can’t say I’m all that surprised. I think Ellison demonstrates a profound lack of understanding for the way open source works in the above interview, so I’m not all that surprised.
Q: An interesting interview indeed. Let’s tease some of the individual pieces out, starting with the basics. What are people saying about Oracle’s move?
Q: Opinions are strongly held, as you might expect. Dave Dargo, who started and ran Oracle’s open source program office, is calling BS on the move. Matt Asay, ex-Novell and currently Alfresco, is also seriously displeased with the decision. Canonical’s Mark Shuttleworth, according the 451 Group, is more sanguine but pessimistic on Oracle’s ability to maintain compatibility. Steven J Vaughan-Nichols is positively livid. Dan Farber synthesizes some of the other opinions on the matter here. Haven’t seen too many folks jumping for joy, though I’m sure there are fans of the move out there.
Q: What about you? Is the sky really falling?
A: I’m not quite as pessimistic as some of the above commenters, but I do have concerns about the deal. I think the degree to which this decision augments, rather than undercuts, Red Hat’s role within the enterprise has been underacknowledged. Mark mentioned it, contending that Oracle’s move “further entrenches Red Hat’s position at the centre of the Linux-for-the-enterprise game” and I agree. The compatibility question is actually a facinating one, and represents in my view the single greatest risk to Oracle’s new business.
All that said, however, Oracle seems intent on leveraging its high margin closed source businesses to suck the oxygen out of an open source market by taking advantage of the model. I’m less convinced than Matt about how happy Red Hat customers are with the value they’re getting for their money – see the Dell/Red Hat pricing tiff from a few years ago or the comment from Henning Heinz on my post here – but neither am I entirely comfortable with the approach taken here.
Q: Let’s come back to the compatibility question in a bit. For now, Vaughan-Nichols says that no one saw this coming – is there precedent for this?
A: From a technical perspective, certainly. CentOS does this already, distributing Red Hat Enterprise Linux (RHEL) for free. CentOS, in fact, is the OS underlying my personal Asterisk implementation (it being what TrixBox is based on), so I’m essentially a free-riding RHEL customer right now.
But is there precedent for the scale, the magnitude of the corporate jujitsu that leveraged open source to undercut Red Hat? I can’t think of one off the top of my head. The closest thing I can come up with is the forked version of Java that Microsoft maintained before eventually forking over a few billion dollars to Sun, but that’s still apples to oranges.
Q: In the short term, what are the implications for Red Hat?
A: Well, Red Hat’s share price will undoubtedly take a tumble. They’re off almost 30% at the moment, according to Google Finance. But the more important question is what does this mean to Red Hat’s business, operationally speaking? It seems likely that they will be forced at some point to adjust their pricing, which in term impacts their margins – the single most important factor in their ability to grow significantly faster than the market average (28.05% to 13.95%). That’s bad news for the Linux marketshare leader, no doubt. Potentially worse news is that there clear dominance in the market place has come into question, virtually overnight. Not in a dramatic sense – Red Hat will remain the dominant player in the short term, but ISVs will undoubtedly begin to play the Oracle card in partnership deals with Red Hat. Put differently, imagine if tomorrow there magically two Microsofts. Think what ISVs could (and would) do with that situation, and you have some idea of what might transpire here.
Q: How about over the longer term? What might we expect?
A: I’m not in position to speculate about the low level impacts of the inevitable pricing pressure, though I do worry that interesting projects like Mugshot might be impacted by operational cost cutting. Suffice to say that I expect Red Hat to be negatively impacted operationally longer term.
That said, there are clear benefits to Oracle’s move for Red Hat. This essentially can be viewed as an acknowledgement that Red Hat is, for all intents and purposes, enterprise Linux. It also – provided that compatibility is maintained – potentially makes Red Hat’s ISV story even stroner. Oracle will also reportedly be distributing fixes back to Red Hat, which in theory lowers their development costs while improving the product.
I’d also expect that longer term Red Hat will explore partnerships and/or acquisitions that allow them to strike back at Oracle where they are weak.
Q: We’ll get to the partnership angle shortly; how about for Oracle? What does this mean in the short and long term?
A: In simple terms, they gain a new business with very little capital expenditure (relatively speaking, of course) and in theory siphon off a portion of the revenue currently directed at Red Hat into their own pockets. I’ve seen it argued that they gain control over the stack – preventing Red Hat from becoming another Microsoft – but I don’t buy that unless they’re willing to break compatibility. As long as they maintain ABI/API compatibility with RHEL, I don’t see how they’re in control from a technical perspective. They’re not beholden from an account perspective, it’s true, and maybe that’s the point. But at what cost? The cost of maintaining and supporting a business in which they have little experience. Operating systems, after all, are something that Oracle has not tried in any serious way previously.
It will also be worth watching community reactions to the move; Oracle, in my view, runs the risk of antagonizing developers against them. It’s not that Red Hat is a singularly beloved corporate entity to that audience – much like Amazon is not – but rather that this indicates how ruthlessly Oracle will exploit (in Ellion’s own words) the community and the open source model itself. From a PR perspective, that’s a risk, and a serious one. The initial wave of reactions described above confirms this.
Q: From a business perspective, why would Oracle choose to get into operating systems? The costs seem to be substantial, and the risk from a marketplace and perception perspective are no less.
A: I think the answer comes back to one thing: subscription based revenue. Witness Ellison’s comment from the FT interview linked above:
We make more money selling software-as-a-service than we make just selling software. I’d much rather be in the monthly service charge business, I’ve said this repeatedly. [At present] a huge percentage of our sales are done in the last week of the quarter: all of that goes away, it’s a much better business model.
The decision to enter the operating system game, to me, seems to be as much about not leaving recurring services revenue on the table as it is about heading off any Microsoft ambitions – which weren’t realistic in my view anyway. As I alluded to in my network offering piece, I see substantial future revenue in layering services around the operating system. The fact that Oracle is replicating not just Red Hat’s code but their network offering indicates that they agree.
Q: How about from a technical perspective? Why did Oracle choose to leverage Red Hat’s code rather than create its own distribution or certify to a competitor like Ubuntu?
A: Well, creating its own distribution in my view would have been a move destined to fail. An operating system flavor supported by no vendors and backed by no community would be more or less finished before it hit the shelves. I never really believed that Oracle would go that route.
As for why not Ubuntu, I don’t know if coming to terms with Canonical was difficult or not, but by going with Red Hat they have one huge advantage – at least in theory. Compatibility. Throw in the fact that they hurt a vendor that via acquisition (JBoss) is increasingly ambitious and threatening to their core business, and it was probably a relatively simple decision.
Q: Let’s return to the partnership angles here; what alliances do you expect to see this event generat?
A: There are too many possible permutations to cover here, but Farber tells us that Goldman Sachs’ Rich Sherlund expects Red Hat to court the likes of Ingres or MySQL, and I’d throw in an even more obvious candidate – EnterpriseDB, a Postgres based product that delivers Oracle compatibility. The best possible return of fire for Red Hat would be to hurt Oracle’s flagship database business. Some have contended that Oracle has used Red Hat to highlight the inherent vulnerability of the open source model, but I think it’s equally possible that open source returns the favor. The throttle there, of course, is partner readiness. MySQL has historically kept its focus on newer, high volume markets rather than trying to going aggressively after Oracle where it’s strongest (although whether the recent Enterprise decision changes that will be interesting to watch), and while EnterpriseDB has some notable Oracle kickouts on its resume they’re a long way from being able to go toe to toe with the Redwood Shores giant. There’s also enterprise comfort levels to consider; as I relate here a CIO panel featuring participants from the likes of Citistreet and Fidelity at a recent OSBC were almost universal in their admission that the mission critical relational database market was the one where they were least comfortable migrating to open source.
Q: Ok, how about explaining the compatibility issue further?
A: This, to me, is the key to the entire deal. To recap, Oracle in its announcement promises full binary compatibility to RHEL. The relevant section in their FAQ (PDF warning) says:
You can continue to use any ISV application that has been certified for use with RHEL3 and RHEL4. Oracle will offer support for the operating system running underneath applications that have been certified with these versions of Red Hat Enterprise Linux.
Translated, that’s Oracle saying, if it’s RHEL 3 or 4 certified, we’ll run it. Red Hat, however, contends that there will be issues:
Q: Oracle says their Linux support includes the same software compatibility and ISV certifications of Red Hat Enterprise Linux. Is this true?
A: No. Oracle has stated they will make changes to the code independently of Red Hat. These changes will not be tested during Red Hat’s software testing and certification process, and may cause unexpected behavior. Hence Red Hat software certifications are invalidated.
Translation: they’re going to be making changes independent of our codebase, so all bets are off.
The truth is that a.) only Oracle can decide it’s supporting and what it’s not, and b.) the ISVs, not the OS vendors, are the ones that will be deciding what platform to write to. In other words, if an ISV chooses to write to RHEL4, and Oracle makes changes to RHEL4 but promises to support it, that’s Oracle’s problem to solve provided that they can get customers to trust them.
The interesting wildcard to all of this is something Donnie brought up in #redmonk, speculating on the possibility that the LSB certification could undermine Red Hat’s claims of incompatibility. Great point, and wouldn’t you know it, Oracle just joined the LSB as a full Platinum member – despite having no prior involvement with the organization. The timing, I think we can all agree, is not a coincidence.
Last, it’s worth discussing the possibility of a fork. Will Oracle, as Mark says, “get frustrated with supporting someone else’s codebase?” Or will they be content to continue distributing a nearly identical version of someone else’s operating system? Tough to say.
Q: Do any customers really want to purchase an entire stack, database, operating system, application platform, and applications from a single vendor?
A: Great question. I have my doubts, Oracle obviously does not.
Q: Who else does this impact?
A: It’ll ripple through virtually the entire commerical open source ecosystem, and reverberate down through a variety of non-commercial communities. In the short term, I expect Linux competitors such as Microsoft and Sun to make some initial headway due to the haze of Fear, Uncertainty and Doubt stirred up by this announcement, but I think that should dissipate quickly as enterprises recognize that this is merely greater investment in the Linux ecosystem, not less. Likewise, IBM is now faced with some very interesting decisions to make with respect to its operating system strategy, concerned as it must be with Oracle’s increasing weight from a critical mass perspective. Novell and Ubuntu must also now contend with a reinforced RHEL = enterprise perception, which will be less than ideal; not to mention a far more cost competitive version of the same bits.
Q: Can we expect to see Oracle move in similar directions in other businesses?
A: According to Oracle, the answer is apparently yes. But I think those expectations need to be tempered by considerations of model. For example, one person I was chatting with recently speculated on whether or not MySQL could be successfully forked in similar fashion. While this would be technically possible for a vendor like Oracle, as MySQL’s code is open source and available, the dual license nature of the firm essentially means that all of the core MySQL developers work for MySQL. So they’d have to not only coopt the code – which is easy – but pick off people – which would presumably be significantly harder. It’s worth mentioning, however, that according to Matt Oracle has been poaching Novell kernel folks over the past few months, obviously in preparation for this move.
Q: Anything else to add?
A: I’m sure that’s quite long enough. For questions that I didn’t answer, feel free to drop over to #redmonk and clients, feel free to IM us or pick up the phone.