It's a seismic shift, this proposed acquisition of JBoss by Red Hat. The open source code/commercial support business model is no longer just riding a commoditizing wave among certain IT stack runtime and tools components. The model is now carving out a disruptive place on a full, modern IT stack basis that extends down to the multi-core silicon threads and right on up to the vanguard of SOA.
For those predominantly commercial source Lock-in is out. Hand-holding is in. (IBM prefers "private" source) vendors, which have made amends with open source and cherry-picked among products to "open," the continued preservation of their commercial code is now suspect. The escalating JBoss-Red Hat stack strikes at the heart of the data center software infrastructure business. This expanded Red Hat stack is both wide and deep. If end users determine that scale, reliability and manageability of this stack meets their needs, this puts significant downward price pressure on the other infrastructure vendors as they sell a solution, or one-throat-to-choke, value.
Indeed, there are two major trends in open source, as evidenced by last week's LinuxWorld show: open source is moving downstream from Linux into virtualization in a big way, and simultaneously open source is moving up the stack abstraction to SOA. In the meantime, open source is hollowing out the middleware components beyond runtime to groupware, portal, directory, and transactional messaging. Tools is a done deal.
Already in the nascent SOA components market, there are open source ESB projects. There is talk of open source UDDI repositories. Process management runtimes and tools can't be far behind. JBoss's JEMS, with a helping hand from HP, alone has made inroads into open source SOA functionality before the commercial definitions of those components have matured appreciably. In effect, the open source model in a few short years has re-emerged into the domain of forging the definitions and reference platforms of the new generations: on IT's forefront of form and function, based on a consensus model (somewhat gamed) but still community-driven. That is seismic.
Red Hat with JBoss now is in a position to solidify the open source and commercial open source models, as well as break out and extend the model into the hottest areas of IT. With the AOL/iPlanet/Netscape assets it picked up 18 months ago and now JBoss, Red Hat has a powerful and comprehensive open source stack for datacenter, edge computing, early SOA, Enterprise Java, Web 2.0, and tools/Eclipse. There are plenty of other open source projects to augment this march, and the clout of the Red Hat/JBoss market share can help make more of those nascent projects gain traction and therefore be successful in community support.
Open source, therefore, is no longer on the commodization continuum, of repaving proprietary cowpaths with less expensive and open alternatives. No, what the JBoss/Red Hat combination points to is the open source model going to the cutting-edge continuum of IT, and offering the simpler, crawl-walk-run path to higher productivity, exploitation of the most modern mainstream IT trends, and then building out those capabilities into the mixed commercial-open source model that JBoss and Red Hat have done so well by. Users do seem to like the model.
The conventional wisdom has been that the computing world is better served by as many as three, but not more than four, and as few as two, full stack providers. This keeps balance, choice, and competition in place, and prevents a repeat of the IBM mainframe hegemony of the 1970s-1980s, and the WinTel duopology of the 1990s. However, with a viable, full open source soup-to-nuts stack, that equation is set askew. The competition shifts from an emphasis on the code itself to the combination of de facto standard and open implementations to a larger emphasis on field execution and support. The higher differentiation value is therefore on the execution in the field, support, vertical business logic, pure TCO, and the comfort of the long-term relationship with each account on the users' terms.
In other words, lock-in is out. Hand-holding is in. The bully sits on the sidelines, while the helpful tutor gets more of the play-time. Question is, is Marc Fleury a bully or a helpful tutor? I guess it depends on where you sit, inside the sandbox or out on the grass.
Here's a smattering of the more tactical implications from today's M&A bombshell:
- IBM will need to rethink its relationship to Red Hat. Good news for enterprises: the likelihood of an alternative, IBM-stewarded open source stack (SuSE, Geronimo/Gluecode, Apache, Eclipse, etc etc) to counter balance the Red Hat-JBoss combination. This is already well in motion.
- Great news for HP, which has been tight with both JBoss and Red Hat. HP has the hardware and services to ride this acquisition well. HP can partner, and let Red Hat manage the stack, while throwing lots of engineering talent at the development process. Hurd is lucky as hell or ... ? (As my high school wrestling coach used to say, Luck has nothing to do with it.)
- Also good news for the non-IBM global SIs, as they seek to be the best hand-holder in the accounts while exploiting the JBoss/Red Hat stack. Me thinks there will be a margin rout in SIs, however, followed by a wave of consolidation (CSC as harbinger?). Low-cost provider wins. It's nice to own a hardware biz too.
- Good news for Novell as IBM, HP, BEA, Oracle, Sun will want to make sure that SuSE is a strong and vibrant alternative to Red Hat's Linux distros as Red Hat expands up the stack into SOA with JEMS.
- SAP would do well to begin segmenting Netweaver to specialized usage, and put an open source stack like the Red Hat/JBoss one under more of its installations.
- Greenfield Web 2.0-type ISVs and startups have a no-brainer now for an expanded, solution-oriented stack partner in Red Hat.
- Sun sees less of a differentiation from price and completeness of its JES stack. It won't be a full Red Hat stack vs. a full Sun JES stack. Look for IBM to better assume the counter-balance to Red Hat role (with an open source DB2?). It was a nice try, Sun, but timing is everything.
- Oracle can not afford to just dabble in open source, it will need to get serious. The business applications business is the only future. Time to write a new rate card on the middleware. And watch out for what IBM does with DB2.
- BEA is going to have to show that its commercial SOA is so much better than the open source alternatives that the high price is a bargain. Sounds like a niche market that will be under constant price pressure, right from the start.
- Microsoft is the only non-open source infrastructure player. Time to write a new rate card. Better use pencil.
- Now Red Hat should buy Borland IDEs, if not for the tools themselves, for the accounts the installed base represents, if nothing else. Support of all those tools fits in quite nicely with the open runtimes.