During a break at Software 2006, I caught up with Bruce Richardson [below], chief research officer at AMR Research. He cuts through the noise when it comes to enterprise software. He comments on SOA, Oracle, SAP and the vendor/user relationship.
Ray Lane talked about how the software industry needs a big rethink, to which Richardson responded, "It's safe to say you have to rethink. We've been saying that for last ten years--when ERP was running out of steam, when all things had an "e" in front of them, rethinking annual pricing, putting 'service' back in customer service. It's like [Boston's] Big Dig, there's always something to that needs improvement."
SOA has been touted a key theme in transforming the software industry, but customers aren't necessarily clear what the SOA attraction is all about. "Technology providers hope that SOA leads to a buying boom, so everyone wants to have it," Richardson said. "It's a land grab, and [vendors] want to be positioned as a hot platform and lure developers."
"If this were ten or twelve years ago, every company would have been in EAI space, they either get bundled in or disappear. The same thing will happen to standalone SOA providers, such as SOA Software, Infravio and AmberPoint," Richardson said.
He also said that customers aren't pining for SOA. "One Fortune 50 CIO told me that he knows that SOA is inevitable from SAP, but he's not in a race to be first," he said. "SOA is a better model, but think how much easier it would be if you were doing it from scratch, rather than dealing with all the permutations with existing code to manage. SAP, for example, has the challenge of service enabling hundreds of millions of lines of code and most customers have a mix of code from different vendors. IBM went from 16,000 applications to 4,000, and has to think about Web service enabling them.
Look at the size of that task--it's easier to put a man on the moon. You have to start with point solutions, such as employee facing services, provision new employees, and then think about getting closer to customers and suppliers." The notion of mass replacement of software isn't appealing. Nor is the notion that all the maintenance revenue enterprise customers pay is going to build something they don't want anytime soon or ever. "They are asking what they get for maintainance dollars outside of a kind of life insurance," Richardson said.
On the notion that SOA and upgrading to a new architecture and enterprise applications is a near term win, Richardson said no way. "There's really not a lot going on for SAP and Oracle. They have to finish their applications. Normally when have new product, no one wants release 1.0. It takes three releases to have the level of functionality that it replaces. Releases occur every 12 to 18 months. Customers have to figure out where it's used, what to integrate, need a year to plan and a year to put it in--so broad customer deployments won't happen until 2012," Richardson said. Microsoft Vista has taken about five years to gestate, and enterprises won't be pushing hard to deploy it until all the kinks are worked out.
He also thinks that SAP has challenges in reaching its stated goal of getting to 100,000 customers by 2010 through organic growth, which means have to add 15,000 per year. "I've talked to SAP's largest customers, and they complain about skill shortage at high end. They aren't getting good answers about SOA, planning for Mendocino (SAP/Microsoft integration) or other issues. Turning the crank seems like a prescription for mayhem."
Will customers with hundreds of millions or billions of dollars sunk into ERP applications move from Oracle to SAP or vice versa? "It's a non-starter. Enterprises don't want to maintain two architectures or go through the disruption." Oracle and SAP say they are halfway done with their SOA-centric applications, but that's like being halfway up Mt. Everest, Richardson said. "I go back the book by E.F. Schumacher, Small is Beautiful: Economics as if People Mattered," he added. "It's the idea of using the appropriate technology. Applications have been over-engineered, and as a result they are towing the Mayflower and 400 hundred years of history behind." Most companies won't bulldoze their legacy applications, but they wish they could...