Some call it IT "sprawl," and there's plenty of agreement that there's too much "clutter" in our enterprises. Is it wise to potentially add new infrastructure via SOA on top of what may already be a tangled mess of applications and systems?
As reported by Brian Sommer in a new post over at Enterprise Irregulars, Steve Mills, IBM's Senior VP of Software, told attendees at a recent Merrill Lynch confab that that today’s CIOs -- at least the ones at large corporations -- are “more interested in the 20-30 years of accumulated infrastructure than the next new thing”.Needed: 'a garbage collection service, not integration technologies that tie one piece of trash to another'
Brian runs with Mills' statements and adds the following perspective:
"SOA platforms are not important. Clutter control is. Some IT application portfolios of F500 firms still look like something a person with a compulsive collection fetish would have. Even with all of the rationalization initiatives of the last few years, F500 firms have accumulated a lot of application rubbish via mergers, acquisitions and organic growth. They need a garbage collection service, not integration technologies that tie one piece of trash to another."
Shareholders and other constituencies want to see organizations do all they can to at least hold the line on IT infrastructure costs, Brian observes.
Fair enough, but much of the "trash" and "rubbish" Sommers refers to are perfectly functioning legacy applications that have been around for up to 20 years. The reason they are "legacy" applications is because they have survived the test of time, and have been upgraded and improved upon countless times. For many businesses, its a better bet to stick with what works than risk it all on some new-age infrastructure that could result in months, if not years, of additional consulting fees and end-user rage.
One of the most compelling value propositions around SOA is that the architecture offers ways to surface various legacy functions as services that will fit in better with evolving new scenarios. I think the discussion needs to revolve around ways to separate the wheat from the chaff. We need ways to determine how to best identify the legacy systems are delivering the greatest value, versus those that may be merely using up electricity and taking up space.