Amid the chatter on the next round of SOA acquisitions, where analysts like fellow ZDNet blogger Joe McKendrick are predicting that Oracle and perhaps SAP will be on a Fall spending spree, I have to agree to disagree. Oracle may well continue to enter vertical and specialized markets with outright purchases of industry-specific applications and data design concerns, but buying big chunks of SOA infrastructure -- they will opt to build instead.
Same with SAP. I imagine that SOA policy-based governance mechanisms are in the final stages of production within both SAP and Oracle, and that they play off of the existing infrastructure strengths for each. You should also expect the same from IBM, quite soon. Policy, directory, meta data ... you want these close to the vest and tightly integrated into your legacy product offerings. Money can't buy that.
Remember, the tick-tock of the SOA bulk-up clock affects those weak in holistic infrastructure -- where they only have one or two major parts and need to quickly acquire the rest to remain in the SOA platform game. This time-urgency is not an issue for SAP, Oracle, IBM, BEA, or Microsoft. In fact these players know from what's going on in their accounts that the demand for SOA governance is just a trickle on its way to a future complexity-driven wave.
Quite true: No one has the complete SOA lifecycle offering. And how could they, the definition of a full SOA lifecycle offering remains a work in progress? But if you are a vendor that already has a major beach-head of applications and datacenter infrastructure in place in the global 2000, then your build-out to SOA will be iterative, balanced, and built upon the fulcrum of your existing intellectual property. As long as the open source SOA projects drag along on a predictable pace, there is no need to buy SOA infrastructure now. (That's why the major vendors have their liaisons/spies in these projects, eh?)
For those that seek a SOA infrastructure future yet lack the big footprint in enterprise IT -- well that's where time is of the essence. Better buy or be bought post haste -- for the tenure of the best-of-breed approach to SOA infrastructure components has the shelf life of a cut pumpkin in the Indian Summer sun.
By the way, want to get into a rip-roaring discussion these days on SOA? Debate the merits/demerits of best-of-breed and holistic platform with the lesser vendors. Hint: best-of-breed wins in design time, but ultimately loses in run time.
Most vendors do expect that SOA scalability demands will necessitate entirely new kinds of management. Triangulating today when in the future the practical need spurs the purchasing and methodological adoption of SOA governance in earnest allows plenty of more time for SAP and Oracle to build these functions into their existing wares.
More likely to be buying SOA infrastructure in the near term, however, are HP (management 'r us), Novell (one more last hurrah!), Sun (software 'r us again), Borland (design time meets runtime), and Sybase (SOA meets mobile). Others to keep an eye on from the buying side include TIBCO and Cisco. You know who the sellers are.
So the other analysts had the notion right. Yes, there will be more SOA M&As activities, but it won't be infrastructure for Oracle and SAP.