1. Apr 11th, 2007

    How will Web Services work with a Web of Services?

    Ismael proposed an interesting test yesterday. Suppose that the Web of Services really takes off. Not in the sense of selling more passes to conferences, issuing hourly press releases, or banking on product upgrades. Take off with businesses actually looking at services they can reuse based on availability, functionality and associated policies. A world where the semantics of SOA apply in practice.

    In this hypothetical world of reuse and composability, people are going to turn towards that which works, is easily reused and commercially viable, so they can move on to solve higher level problems using existing solutions. They’re going to rely heavily on services providers, the SalesForce, Amazon and Google of this world.

    I did say it was hypothetical.

    But bear with me. So now we’re in this twilight zone where developers spend more time reusing existing services than building new ones. Practicality happens often, and outweighs any discussions over the injustices of WS-ResourceTransfer. The most transferrable skills, the best return for your code, comes not from where vendors are pointing you, but where existing service providers are.

    And there lies the test. Imagine, if you will, that world of rampant reuse. Of a Web of Services that crosses networks, applications and use cases. In your opinion, which Web technologies are more likely to mature and survive?

    1. Apr 11th, 2007

      Phillip Toland

      The web services that are most likely to mature and survive are the simplest ones. The more complex a service is the more difficult it is to use, particularly as it crosses organizational boundaries.

    2. Apr 12th, 2007

      Sterling Camden

      Similar to Phillip’s “simple” answer, but different: those that are most atomic. Use what you need, with no excess baggage or requirements. Mix and match freely.

      But I think you were talking about the underlying technology rather than the structure of the service itself. In that case, I predict a successor to anything we’ve seen so far — built on existing technology, but simplifying its usage. I have no idea how, at the moment.

    Your comment, here ⇓

    Or using OpenID