Tuesday, May 16, 2006

SOAP vs. REST Pointless?

Steve Jones thinks that SOAP vs. REST is pointless. I assume from his post that he thinks that POX is pointless too.

I can understand that opinion. I think everyone in IT integration is frustrated that every wave that comes by that promises to give us massive reuse fails.

The good news is that IMHO services are here to stay. At the very least, they will get a fair shake. Especially with success stories like this from Amazon.

Regardless if Steve thinks the arguments are pointless or not, the WS-* standards aren't there yet. And SOAP really doesn't buy you all that much out of the box. Sure maybe if all the higher level standards amount to something it might, but I'm not holding my breath.

I think that dissent is important.

Steve argues as Anne Thomas Manes does that the tools will save us. He even cites XDoclet as an example. Funny that I use that as an example on why tools are not the answer.

Like I said, the good news is that services are here to stay. How about we just go back to the drawing board on some of the things that are broken instead of slogging on with the 60+ specs? I don't see that happening. There are too many vendor-pires too heavily invested at this point.

The only answer is enough simple OSS rebel frameworks and hopefully enlightened vendors that treat POX, REST, and whatever innovative ideas come next as first class citizens. The truth is there is massive premature standardization occurring. We might get there some day, but you can't just beat people into submission. IT Integration isn't the commodity that Steve describes. At least not where I work.

A big complaint I have against drinking the SOAP/WS-* Kool-Aid is I haven't seen any real success stories on following the WS-* approach with heavy integration. There are many around SOA, but strangely the specifics of what is on the wire are often left out. Admittedly I may be missing these - please let me know if you know of any.

In general I think that the concepts of SOA and EDA are what are really important today. If you can break things down at the proper level of granularity and use whatever you want on the wire (try a couple!), you will be fine in the end. People including me in the past do a lot of hand waving about standards, service contracts, tool support and what not which are important - there just not as important as you think - at least today.


markg said...

I think you are right on with this. Far to many folks think IT Integration is easier than it is. I think vendors and analyst tend to fuel this. The hard part comes when managers listen to it and then expect the parting of the Red Sea.

Anonymous said...

"There are many around SOA, but strangely the specifics of what is on the wire are often left out."

WS has only one purpose: to turn developper into enablers and lock-in customer into their proprietary API.


The chutzpa of WS promotors is that they advertise the very crippling "feature" as a near magical cure.

The only rational logic behind it is rather simple: if our richest customers bought-in Java and Windows, why not sell them WS?

Laurent Szyster

fuzzy said...

Preaching to the converted brother.

The evile plan isn't going to work though (man at least I hope not).