- From: Savas Parastatidis <Savas.Parastatidis@newcastle.ac.uk>
- Date: Tue, 4 Nov 2003 19:56:10 -0000
- To: "David Snelling" <d.snelling@fle.fujitsu.com>, "Steve Graham" <sggraham@us.ibm.com>, <public-ws-desc-state@w3.org>
- Cc: <www-ws-desc@w3.org>
Hi David, > > I agree. My hope had been that the full Service Data model could be made > available to a larger community, but this does not seem possible > (sensible) > through the WSDL layer. A higher level should be able to meet the dual > goals > of simplicity and power that Service Data represents in the Open Grid > Services Infrastructure (OGSI). > You know only too well that for the last few months Jim Webber and I have been arguing against the perceived requirement that SDEs are necessary in the Grid community, even more so about the view that they must be introduced in WSDL in one form or the other (e.g., as attributes). Service Data Declarations appear to be powerful only because of the OGSI view of the world, where each resource is exposed as a Grid Service Instance and where consumers are allowed to create or request the destruction of a Grid Service Instance. My limited understanding of WSA tells me that seeing each resource as a Grid Service Instance, which is defined as a potentially transient, stateful Web Service with lifetime management related characteristics, is not consistent with the Service-Oriented model but, rather, with an object-oriented model similar to CORBA and DCOM. If the Grid community wishes to define separate specifications that add additional characteristics like Service Data Elements, factories of resources (it's my understanding that we are moving away from the term Grid Service Instance in favour for "Resource"), they are free to do so. However, I am worried that such an approach will further distance the Grid from the world of Web Services without actually that being necessary. Best regards, .savas.
Received on Tuesday, 4 November 2003 14:58:06 UTC