More on REST vs. WS-*

There is an interesting debate going on between my friends Ted Neward and David Chappell about REST vs. WS-* complexity. Beyond the political analogy (in which I agreed with Ted) I would like to add my own opinion on the topic.

One of the main attractive of REST as an ARCHITECTURE STYLE is its simplicity compared with the traditional SOAP-based approach. Simplicity is a great thing but it just can’t be applied to all scenarios.  Aspects like Reliability, Security, Federated Identity, Transactions, etc, represent very complex challenges of SOA solutions. And as we might already know, complex challenges can’t always be addressed with simple solutions. Approaching those challenges with a REST type approach will end up (in the best case) on recreating a REST-based version of the WS-* protocols. I think the whole WADL vs. WSDL 2.0 debate is a good example of my point. Secondly, WS-* protocols always propose transport-agnostic (a.k.a NON HTTP DEPENDENT) solutions to those challenges. Although I agree 100% with my friend Harry Pierson in that REST is an ARCHITECTURE STYLE; the fact of the matter is that almost all the REST-based solutions I’ve seen for addressing things such as Reliability, Security, etc are dependent on HTTP as a transport protocol and consequently are not a viable solution for REST services that don’t use HTTP as a transport which we have plenty of those.

 Some of the most talented people in the distributed systems industry has put a lot of time and effort creating the WS-* protocols. Not taking advantage of those is, in my opinion, just denying all the progress the industry has made trying to build interoperable Standards for addressing some of the top challenges of Service Orientation.