Hey guys, who but marketing said that SOA requires anything like SOAP RPC? It is the most primitive and speculative approach to use SOAP RPC and call it SOA. If we talk about Web Services, I care about WSDL only. SOAP or REST are just transport/protocol models for SOA, which can easily stand on IIOP, for example. A crowd of RPC or message calls is as far from SOA as relational model far from OO.

To me, the easier protocol the better. However, please, make it standard - mandatory for all. If you can do it with REST, I am for it. Otherwise, sorry…

Thanks, Andrew at Digital Earth Weblog for today’s entry. I’ve been noticing lately a bit of a discussion about REST vs. WS-* that’s been going on.

Every now and then some voice of reason raises up above the hype hypnotised crowd and offers the chance for awakening to the ones that have not been completely brain washed.

I’ve done both styles of AC systems on multi-million even a billion dollar project and I am yet to see where
simple REST is less useful than standardize-everything, interoperability-with-everything WS-* stuff.

The biggest champions of WS-* are the vendors like IBM who sold you some billion dollars big iron thirty years ago and now instead of helping you to migrate to a more nimble, future proof platform they are selling you a
‘pig on a lipstick’ WS-* friendly, ESB solutions that will make your COBOL/CICS based iron run forever.

Just buy 100 million WebSphere ESB bus and everything will be just fine … Just keep putting wrapper, over wrapper …

REST in this case is really selling you a different model. Instead of selling you a SOAPey story of ultra-interoperability and standardization it is giving you the value what good old webservices
were meant to provide you. Location independence, simplicity, universality.

I do not care if eBay’s webservices are WS-* compliant as long as they are easy to learn, easy to use, performant and if they add value to my electronic business.

Isn’t that the formula behind every technology that successfully took off.

It seems that the REST advocates believe that if they keep repeating it over and over that WS-* is astronomically expensive and requires too much fiddling, than it will some how be true. Sure, some vendors and approaches are too expensive and sloppy. The same can be said of a REST based solution! I can only say that this satiric parable is completely contrary to my experience in how I (and my customers) deal with WS-* based systems, and the success rates with them. People are doing good work with this technology, and it’s yielding tremendous benefits.

These “my tech is better than your tech” pissing matches are not an effective way to promote a philosophy’s merit — it’s a form of mockery and self-congratulation. Perhaps that’s your point, and if that’s the case, great, we all had a laugh. But I think you’e actually making it harder for people to buy into this approach because of the attitude. You’re falling into an evangelical trap that will actually hinder adoption, though perhaps that’s part of the goal, to stay an elite club.

I love the REST style and think Roy’s thesis is one of the most clear and illuminating works on architecture in years, and I would love to bridge the gaps among the skeptics (which, in my customers, are near 100%+) where I can, but this kind of stuff doesn’t help.