Perhaps every software application from the era of punch cards to today's service-oriented architecture (SOA) started out with good intentions, but the road to cost over-runs is paved with good intentions.
Good intentions are no substitute for a long-term application strategy, according to R "Ray" Wang, principal analyst, Forrester Research, Inc. Before offering guidance on how to do it right, he offers a few application horror stories in "Why You Need a Long-Term Apps Strategy," which is the first in a series of upcoming Forrester reports on "Building A Long-Term Apps Strategy."
While not naming names, Wang points to a "big oil" company that tried going the single vendor route for all its packaged applications. Unfortunately, after delivering human resources and finance applications, the single (also unnamed) vendor stumbled on providing the additional apps the oil company required.
Things got worse when oil company executives and managers finally realized that single vendor wasn't going to work and switched back to best-of-breed package applications. It turned out that the middleware from the vendor of choice was not adequate and fixing that problem alone cost $11 million.
Best of breed is not always a panacea either, as Wang points out in the case of an un-named retailer that selected a series packaged applications without first determining how they were going to be integrated.
In that case Wang writes: "The lack of an overall apps strategy or vendor standardization may result in an increase of up to $20 million for integration costs, maintenance fees and vendor management expenses over a three-year period and significant long-term upgrade costs."
These scenarios are the result of too much piecemeal decision making aimed at solving short term problems as when an executive says: "We need to upgrade the accounting software. What's available?"
Avoiding these horror stories while moving to the next generation of applications and architecture requires long-term planning, Wang argues. This is especially critical in an SOA era as organizations are entering an upgrade cycle for packaged applications, such as human resources management, that were last implemented as part of the Y2K transition.
"As the industry moves to SOA and the beginning of the next upgrade cycle, business process and applications professionals should take advantage of this unique opportunity to plan strategically and initiate a long-term applications strategy," Wang writes.
Asked about the role of SOA in this planning process, Wang replied: "SOA is definitely a key component of the 'Long Term Apps Strategy' as well as part of a larger architectural view. As enterprises standardize deployment around a few core vendors, packaged applications provide one major gateway on the journey toward SOA. Next-generation packaged applications promise to use SOA designs and Web services to let customers tailor their applications software to their specific business needs."
Packaged application vendors are converging on what Wang terms "four major middleware ecosystems," which he has identified as IBM WebSphere, Microsoft .NET, Oracle Fusion middleware and SAP NetWeaver.
A major task for enterprises planning a long-term apps strategy is to decide how many middleware vendors it is willing to support, Wang said.
But it goes beyond that to the big picture of "technology strategy," which the Forrester analyst identifies as including having a clear middleware and SOA strategy, as well as plans for upgrade planning, master data management, deployment analysis and security.
The good news, Wang notes is that packaged application vendors are configuring their software for the Web services and SOA approach, so the upgrade cycle offers an opportunity not only to improve functionality, but architecture as well.
"As vendors design with SOA in mind and build componentized Web services, each new release offers a more compelling case for upgrade or replacement," Wang writes.