In a response to my post on Phil Wainewright's post (I think I caught all the hops there), Ian Bruce points to a new article that talks about the success Starwood Hotels is seeing with its SOA strategy.
The hotel chain expects to be fully SOA enabled by next year, and expects to save at least $20 million a year by dismantling its mainframe, a dismantlement that is possible because of moving its CRM and online reservations applications off Big Iron onto standardized, commodity-based platforms.
Miko Matsumura also responded to my post with a pointer to a recent article detailing Sprint's SOA efforts. Sprint is keeping its mainframe, "encapsulating mainframe applications as 'virtual services' within EJB (Enterprise JavaBeans) wrappers that expose the functionality of the applications using SOAP, WSDL (Web Services Description Language), and XML Schema." Sprint no longer "has to reinvent the wheel" every time a new customer is brought into the system, the article says.
Interesting -- one company leverages SOA by throwing away the mainframe, while the other keeps the mainframe at the core of its strategy. That just shows you that SOA is going to evolve in many different forms. No two SOA strategies will ever be alike, and there is no 'right' or 'wrong' way to move to SOA.