I've been asked a number of times whether SOAP is becoming obsolete. Cited examples:
- Amazon offers both SOAP and REST interfaces to its API. According to Amazon, REST was used 85% of the time.
- Google quietly dropped its SOAP API in late 2006.
- Ruby on Rails 2.0 will drop its SOAP API in favor of a built-in native REST API.
So does this signify the end of SOAP?
Not at all. It's a matter of SOAP recovering from its initial overzealous hype. According to Gartner's Hype Cycle, SOAP would be leaving the "Peak of Inflated Expectations" and, as the above examples illustrate, entering the "Trough of Disillusionment."
For simple web-based, language-neutral APIs and for point-to-point communication, REST is simpler. It has a lower barrier of entry, smaller bandwidth requirements, and is better for AJAX and modern web development techniques. SOAP was not designed for this type of environment, and here REST is definitely a better fit.
It is worth noting that SOAP originally stood for Simple Object Access Protocol. This was found to be misleading: SOAP implementations tend not to be all that simple. Therefore, this acronym was officially dropped. It now stands for nothing. (On a related note, here is a humorous fictional conversation about SOAP: The S stands for Simple.)
However, where SOAP will make an impact is in large-scale Service-Oriented Architecture (SOA) implementations. It is a standard that is replacing proprietary EDI implementations, and has pretty much killed CORBA/IIOP. Companies such as IBM are building SOA infrastructures and helping guide the standards to make it a smoother experience for others who will be doing the same.
Also consider this: The US DoD is preparing to roll out a massive joint-service SOA implementation using SOAP and UDDI. SOAP is mandated by the DoD IT Standards Registry (DISR) and is a big part of the DoD Net-Centric Enterprise Services (NCES) initiative. SOAP-based web services are built into the latest version of the DoD Architecture Framework (DoDAF) version 1.5, and expanded guidance is likely to appear in the forthcoming DoDAF 2.0.
So SOAP is not dying, and in fact I believe the contrary: it is poised for takeoff (or the "Slope of Enlightenment" in Gartner Hype Cycle lingo).
3 comments ↓
I think SOAP has its place. The problem is that tool vendors promote RPC over SOAP approaches and RESTful web services promote better practices such as document orientation, statelessness etc. (not that everything is rosy with REST but that’s another story :))
The corporate world has never heard of REST and is just now gearing up to adopt and deploy the latest “SOA” web services fad for enterprise architecture. SOAP is an essential component of SOA. Indeed, Microsoft has already pretty much adopted it for .Net services.
REST still has major problems to solve to be viable for enterprise systems. It’s great for open internet applications where a large set of consumers will help tool the specific application. The REST communities cold shoulder to service descriptions (WSDL equivalents) and robust communication standards (WS-* equivalents) are serious problems for enterprise adoption. Both are solvable, but it doesn’t look like the REST community wants to solve them, and they often (wrongly) deride them as unnecessary.
Leave a Comment