JP Morgenthal, who has been pouring healthy doses of reality checks into the SOA zeitgeist for a number of years now, says he has been seeing the beginnings of a disturbing trend as of late: developers and users alike stuck with an irretractable service, in which interfaces and other services have grown deep roots.
JP calls this the "Facebook Effect." Essentially, while comparable or better alternatives may exist for various services on the online market, "Facebook, as a service, is seemingly entrenched to a point where it cannot be unseated." The net result, as JP puts it:
"The user has invested in customizing the service to a point where it is extremely painful to recreate in a completely separate service."
As the service foundation within an enterprise matures and gains traction, "users will start to invest their time in building connections with and automating their processes around the service. This will greatly limit enterprise IT’s ability to arbitrarily change the service in a way that impacts the user."
Building a highly adaptable shared-service layer -- in which any and all interfaces and services can be swapped out on a moment's notice -- has always been the ultimate ideal of SOA. Of course, most organizations, saddled with various vendors' platforms and legacy and post-legacy systems, aren't there yet. And, as outside-the-firewall cloud-based services get mixed into the equation, lock-in -- be it with a vendor or with a particular class of services -- becomes even more of an issue. Many cloud engagements, in fact, may be easy to get into, but incredibly messy and expensive to back out of.
With most Facebook apps, it's just an inconvenience for the consumer to make a switch, with no real money lost. With critical business processes hooked to one service or another, it's a different story. That's why good enterprise architecture and planning really make a difference.