“At first glance, Udi Dahan, Microsoft Solutions Architect MVP, would seem to be the blogosphere’s lone dissenter. I think skepticism is a healthy thing when investigating new technologies, so below are a few of Udi’s comments for your consideration.”

“Udi Dahan is very skeptical about the usefulness and relevance of Astoria and Microsoft’s data service approach in general”

Your question corresponds quite well with the general feeling of confusion I’m picking up in the community. There is so much new stuff coming out technology-wise, but developers are pretty used to that. The things that throws many of them is the new architectural stuff – SOA, REST (now that Microsoft is all over it), Software as a Service, ESB, etc. When all that is topped off by a healthy dollop of Software Factories supporting technology X for architecture Y, well, I don’t know about you, but I’m concerned. So are many development managers who have committed multi-year projects to a technology they have just learned will be obselete in a couple of years.

Anyway, to the question:

REST and SOA deal in a scope that is larger than a single application, so I wouldn’t suggest you use them there. SOAP is just an XML based “protocol” for data exchange. This makes its scope smaller than your application, so, of the three, it is the most suitable.

A better answer would be that if you haven’t run into the need for them, don’t use them. Stand-alone applications shouldn’t be affected much by these issues. Once you go “client-server”, or something like that, I would be looking at them more.

I totally agree with you. Since we can’t count on clients doing the appropriate transaction management, also given that it changes every so often makes older clients a danger, we need to handle things at the DB level. Definitely hell, although the DBAs may be getting some raises out of this

My apologies for using overly flowery and open-to-interpretation syntax. All I meant by it was that you said that Astoria was both new and notable.

Just to be clear, I have not given it any “stamp” of approval. I think it might be a bad idea for the reasons you listed and even thought about that before including it. I just listed it as a list of items to be complete. I do not in any way endorse it and I need to do more research.

I’ve been thinking about something like Astoria for about 12 months, and let me tell you it is a fantastic approach… you should have a read of my posts on Data2.0 which is basically the same topic.

SOAP data services are just stupid… but REST data services? Well that is another matter, the idea of a conceptual model providing URL predicability when combined with the REST idea of verb consistency, perhaps provides the ability to create an intensional language. I.e along the lines of SQL for the web. The key building blocks of SQL are very similar to the building blocks of this idea….