As soon as Web Services started getting buzz I was both excited and concerned. Interoperability and reuse are great things. Shorter time to market is a huge benefit. Bandwidth is limited. That last was never brought up by the sales people taking clients to $1000 dinners while pitching $30,000 web service platforms with $1,000,000 support contracts.
Web services are still a great way to expose legacy systems to myriad clients across the enterprise. Where they become expensive is when they are built with only one or two expected clients to support a (myopic) SOA vision. Especially when many of the new services being built are only aggregations of other services that will generally be a specialized interface to business logic required by a limited number of clients.
If an architecture includes web services, a list of clients must be part of support case or the design is simply buzz word bingo.
This technical rant was prompted by a developer.com article that has way to much code to share with the folks who will often make the final decision, but may get their advisers excited enough to explain it to them.
© Scott S. Nelson
