A long time ago in a galaxy far, far away (well ok, in northernEngland), a group of Rebel scum (i.e. Savas, Paul, Thomas, and me) wrote a paper whose purpose was to demonstrate that all the “special” things that were needed to enable Grid computing could be aciheved using stable Web Services technology (specifically SOAP, WSDL, and WS-Context). We were, unofficially, villified and it seemed that only a few people were prepared to be associated with us in public (in particular Steve Loughran was a gent as were many of the UK e-Science people and their friends).
That paper caused a huge outcry in the Grid community and led to the deprecation of the “special” OGSI set of Grid technologies (not that the GGF has ever acknowledged this officially). The OGSI work was replaced with WS-RF which maintained the same object-like model including all kinds of crazy non-scalable stuff like renewable references (yeah, a WTF indeed!).
While the likes of HP, IBM, and small players like Globus went about happily implementing this set of specs, the rest of the world carried on doing similar things with plain old Web Services and reaped the benefits of widely adopted standards and commoditisation.
As of today the fragmented landscape that has dogged management and eventing with Web Services is beginning to calm down. WS-RF is now effectively deprecated and the 800lb gorillas (including Microsoft who had not lent their support to WS-RF) and their smaller primate partners are moving towards interoperable standards within the W3C.
A friend of mine, who can’t be named for various political reasons, enthused to me this morning, “I told you so.” I agreed :-) Here’s to the Grid and the Emperor’s first set of actual clothes!
Update: Grid computing pioneer Ian Foster points out that moving from WS-RF and friends to WS-Transfer and friends should be a simple matter and that he is delighted with this outcome. I think it will be interesting to see how the Grid community adapts to the third major shift in Grid plumbing in five years.
