To their credit, Red Hat runs a concise analyst event. Three hours of presentations from senior executives and a customer (Qualcomm), some Q&A time, lunch, and you’re out and on to your next engagement. The larger vendors can’t realistically attempt this, but everyone who’s not huge would do well to learn from the open source pure play. You get the big picture, the opportunity to ask a few questions, and the chance to engage in person. Sort of a “No Fluff, Just Stuff” for analysts.
And if that doesn’t convince you, think of it this way: you know they’ve done a great job when I don’t mind getting up at four and putting on a suit to attend.
But of greater concern to most of you would be the substance: what did Red Hat want to tell the mixed crowd of financial and industry analysts? The big picture message from Jim Whitehurst (CEO) and Charlie Peters (CFO) was stability and modest growth in spite of the worst economic climate we’ve seen since the thirties. The last time Red Hat convened this crowd, the market dropped 570 points, which was a record for two whole weeks. A couple of reactionary trends, such as a shift from multi-year support and service contracts to single year commitments, have begun to right themselves, and overall the picture painted by the staff was of a confident, well run organization with tailored performance metrics tempered by an understanding that its installation to revenue ratios will always be biased to the customer relative to its proprietary competition.
From a technology perspective, the content was audience appropriate, which is to say that it was kept at a high level. Financial analysts, after all, can hardly be expected to appreciate the architectural distinctions between, say, KVM and Xen. But below were the key themes on the day.
Fighting the Good Fight Against Lock-In
EVP Products and Technologies Paul Cormier’s presentation differentiated Red Hat from a historical backdrop of proprietary lockin. From the 80′s proprietary stacks of HP, IBM, Sun, and DEC to today’s Windows and VMware virtualization offerings, the same patterns repeated themselves according to Cormier. Just as the open source antipattern has, at the middleware, operating system, and virtualization layers; in the form of Red Hat’s own product offerings, in fact. Though we could debate the minutiae of his case, the fact is that lock-in has, historically, been a popular tactic within the software industry. As the cloud is so adeptly proving at the moment.
Given that Oracle has effectively cloned Red Hat’s primary software product, I think it’s safe to say that Cormier’s claim that Red Hat is open is eminently defensible. The question for me, then, is this: do customers care? Logic would suggest that they would, or at least that they should: who would choose, voluntarily, to be singularly dependent on a vendor? The facts, however, suggest otherwise. As a species, we’re ever willing to trade the future for the present, and the businesses we create reflect this. Whether that means that your average enterprise is continually living Santayana’s Law of Repetitive Consequences or merely that they’re Getting Things Done depends on your philosophy; either way, the practical outcome is the same – businesses lock themselves into technology every day, every hour, every minute. There’s one doing it as I write this sentence.
So while we know, empirically, that lock-in is best avoided, we also know that businesses have, are and will continue to lock themselves into providers, logic be damned. Will customers really advantage Red Hat in a competitive evaluation against, say, Windows because they’re more open? Some will. But the evidence suggests that the market as a whole will not.
That said, there are good reasons for Red Hat to impress upon its analyst followers the importance of a lock-in alternative position. One, because it differentiates the firm from many of its larger competitors. Two, it provides historical perspective: as popular as lock-in might be, it is not an indefinitely viable strategy. Three, when the product is competitive, a certain percentage of customers will pick the more open of two choices. Four, we’re in the midst of a significant shift in technology purchasing and deployment, and lock-in is a more serious possibility than ever. And lastly because lock-in, like its negative image openness, are easy high level messaging tools because they’re binary characterizations. White/black, good/bad, it’s good copy.
If the analysts can understand and appreciate the fact that open source can earn Red Hat the goodwill of developer communities the world over, so much the better. But either way, it sets Red Hat apart.
Parting the Clouds
Because you can’t have an analyst event these days without talking about the cloud, the subject was sprinkled throughout the day’s presentations and discussion. Brian Stevens, CTO, however, got a slot on the agenda to discuss Red Hat’s ambitions. At last year’s event, Stevens discussed briefly his plans to open source the cloud. This time around, Red Hat was able to talk execution alongside of ambition. Between the KVM equipped 5.4 RHEL and API abstraction projects like Deltacloud, Red Hat’s doing quite a bit, project-wise, to drive forward their vision of an open source cloud.
Talk of getting into the cloud hosting business was dismissed; the ambition is purely to be an arms supplier. And given that they supply at least one of the major cloud providers, that ambition seems to be well founded.
The timing for an open source cloud will be important, given the accelerating customer acquisition efforts of proprietary vendors and the increasing volume of PaaS-style abstraction layers that trade the freedom to leave for convenience and time to market. While cloud offerings have to date been constructed largely from open source components, the functionality exposed to users has tended trended away from standardized interfaces. Witness the truckload of quote unquote open and yet highly differentiated APIs. If Red Hat is to help customers stave off a write-once, run-in-one-place future in which applications deployed to one cloud are guaranteed not to work on another, the time is now.
What’s Missing from the Stack
One of the threads running through the day was the idea of an end to end stack. Whether it was a standardized hardware and software combination running at an internal datacenter, or one spun up on an external cloud instance, Red Hat wants to be able to provide a complete and open alternative to its customers, current and future. Via its acquisition of JBoss and its build out of the management capabilities, they’re closer than ever to that goal. But an important question remains: can an end to end stack not include a persistence capability – database or otherwise?
If the scope of the language was less broad, perhaps. But when the big picture strategy is to compete in the stack game from top to bottom with open alternatives, it’s going to be difficult not to play in some portion of the database game, I think. Microsoft and Oracle will certainly leverage their databases competitively, and while VMware lacks a database, it did recently add a framework and management tooling in SpringSource and its Hyperic piece. Consolidation, as has long been anticipated, is afoot, and Red Hat has responded aggressively in every area but databases.
Besides the competitive pressures, the lack of a database offering tends to contradict the uber anti-lock-in messaging. Cormier’s presentation, for example, spoke about lock-in as an endemic problem to the industry, one that negatively impacted customers repeatedly in the operating system, middleware, and virtualization markets. Creating the market opportunity, or one of them, that Red Hat is built upon. Cormier’s macro assertions are pretty reasonable. But the slides noticeably and conveniently omitted one of the largest – and most lock-in susceptible – markets in the industry: relational databases.
As my analyst colleague Gordon Haff covers, the vendor has finally left the management software sidelines, and appears committed to those capabilities via the RHEV-M component. Meaning that Red Hat offers just about every piece of infrastructure software an enterprise could need…except a database. Pressed on a similar question during the day, Whitehurst assured the audience that Red Hat “was not looking to move into new markets,” with a database the given example.
Which is fine for now, because management’s conservative, measured pace has proven itself valuable over time. But as I’ll discuss later, the time may come when Red Hat is compelled by market forces to deliver a true end to end stack: one that includes the ability to persist data. The really interesting question is this: how would they do that? But that’s a post of its own.
Disclosure: Red Hat is not presently a RedMonk client, IBM, Microsoft and Oracle are.