[this is a resend, without the attachment, which apparently made the message large enough that the OGF mail system put it into limbo.] Ian, I think that we're actually in basic agreement! Some comments in line:
That said, I don't see why, if a group emerges "bottom up" with a well-argued and technically strong case for a specification that isn't in the roadmap, we would want to forbid that group from forming. E.g., maybe some comes forward and says that we need to define this particular XAMCL profile, so that we can share this particular set of policies. I doubt that's in the roadmap, but it could well be important to some group.
I agree. We should be aware that the strategy roadmap is going to be out of date the moment it is published, if not before. It will/should encompass the "known" tactical issues, but it can't pretend to encompass the unknowns. If it is *sooo* broad as to do so, it will be too broad to be useful. Put another way: I don't think that the output of the TSC is going to be, or should be, at all tactical in scope. Strategy is our middle name. It's entirely possible that we should first be figuring out what we think a roadmap *is*, and running that idea by the board, before spending a lot of time building one that they don't want. I'm pretty sure that the board is not looking for an OGSA roadmap. It might be looking for a roadmap of roadmaps. The 'grid universe' encompasses multiple communities and multiple vendors - we have to live with multiple approaches. How do we produce a strategy that is general enough to encompass them all, or at least all the ones that are represented in the OGF membership, without it being so vague that it is useless? I don't know. But I think we need to figure it out as the first task.
You express some concerns about OGSA. I think these are important issues to discuss. However, before we do so, we need to be clear about what we mean by the term. Is OGSA:
a) A grand, top-down, boil-the-ocean attempt to define all possible service interfaces needed in current and future grids? Or:
b) A principled approach to define, in an incremental but consistent manner, simple Web Services-based interfaces for the most important things that people need to do in grids, like job submission and management, data movement, etc.?
We've debated this in the past, and I guess you can tell which side I come down on (-:. I firmly believe that (b) is the approach we have to pursue. I also note that it is the approach that we are pursuing in practice.
I have a couple of thoughts about this. First: the fact that some people consider OGSA as a "boil the ocean" solution indicates that there's a pretty serious image problem. That might also make it something that can be positioned as strategic, rather than tactical :-) Second: if it's an attempt to define interfaces, then it's not getting the job done. Alan Yoder might have put it best at the last GGF: "If you gave me a million dollars to go and implement the OGSA data management standards, I wouldn't know what to build." I don't think it's an issue of web services vs. not web services. It's an issue of crisp definitions that lead to implementations in the next 12-18 months. That's what I hear the board, and thus presumably the members, asking for. And I'll say it again: setting a goal to build real operational grids using OGSA based components is not a valuable strategic goal for OGF. Not that OGSA is bad ... but OGF doesn't exist solely for OGSA. Statements along these lines lead to the impression that OGSA is a "boil the ocean" solution. And, by the way, I think that the TSC isn't the right place to define the OGSA roadmap :-) So. As an example of a technology roadmap, I'll offer the SNIA storage roadmap. I'm not convinced that it's a perfect roadmap, or the right roadmap, but it's a concrete example that we can look at critically. Unfortunately, it has not been released publicly. Since OGF and SNIA are partners, we should be able to get a copy of the draft - I'll work on that. I've attached a public presentation about the roadmap, and offer the following quote from the introduction: "The roadmap is intended to help the SNIA achieve two primary objectives: 1. Becoming the trusted authority on information related to storage; and 2. Advancing adoption of storage networks as complete and trusted solutions." What might we do that has similar goals? Mark, since you're listening in, would the board have us meet similar goals, or something different? Best, chris