| |
Comment: |
nd then some more from Charlie, where he quotes himself from another thread:
(yes, I am glad we are having this discussion too!)
Paul, I agree- both of these customer sets - the direct and the indirect- are needed for OGF to be worth doing. I think you have captured these two
distinct groups just right.
I also resonate with your point that some organizations roll their own, and in fact 4 of the software components in my TeraGrid "provider" list (in my
previous message "OGF's primary customers) were developed by, or adapted by, providers I funded. I had not through of it that way, but I agree and
it's great to be able to be a card carrying member of two stakeholder camps! :-)
In my earlier note I said the board would hear about some plans from the eScience ghetto between now and the board meeting. Those plans are moving
pretty fast so let me go ahead and outline them as they are part of this same discussion having gotten feedback from Geoffrey and others that this is
a reasonable summary.
Here is how I have described this plan to the Globus and Condor guys this week (and, as luck would have it, I ran into Rich Wolski - UCSB/
NetworkWeatherService - at the airport and got his take as well):
<Interanal Quote from another thread>
After a very productive discussion about this within the eScience pocket of the GFSG on Monday this week some of us believe that we have begun to
hatch a plan to try to get these software providers re-engaged with OGF (or engaged for the first time). In a nutshell it is as follows:
--- Create a "software forum" at OGF meetings, starting in North Carolina, where any software
--- provider can schedule a room and a chunk of time to meet with the people who use their
--- software. Some groups may want an hour; others may want half a day. The meetings
--- would attempt to get dialog going between providers and their customers, and among
--- their customers. These would take place over 1.5 to 2 days. Organizers of OGF-19 will
--- need to tell us whether to do this on separate days or not so as not to interfere with the
--- standards working sessions.
To design such a forum, we plan to:
a) talk to the major players (Globus Alliance, Condor, Unicore, etc.) to see if they would work with us to create a structure at the OGF meeting in
North Carolina that would enable them to meet with their customers and facilitate their customers meeting with one another. Catlett agreed to talk to
Globus and Condor; Gentsch agreed to talk to Unicore. Anyone who wants to help is free to talk to other providers.
b) make that structure open to *ANY* software provider group who wishes to participate (indeed we want to recruit and advertise- there are lots of
important software providers to this community).
Part (a) is important, because we need some "anchor" tenants to attract critical mass.
OGF, and particularly the standards activities, would benefit from the second-order effects of such a forum. Some of the second-order effects of such
a cluster of "software provider forums," happening in the same venue over ~2 days, would include:
- cross-fertilization between the software providers (the bees being the common users and
the spies sent to see what the other guy's users are saying)
- sharing problems/solutions between users of these packages
- requirements gathering for software providers (while standards work prioritizers listen in)
I also believe that we need to have a discussion between a set of key software provider leaders and TCS to have the following exchange:
a) brief them on the draft roadmap
b) ask them what they believe OGF should focus on in order to help them deliver better
software to help organizations build and operate grids.
I heard three flavors of "questions" about this idea on Monday:
Q1: why would the providers do this? what is in it for them? they have their own meetings
already (globusworld, condor week, etc.).
A1: because putting on a big meeting is a lot of work, and you only get the faithful to attend.
Participating in an OGF cluster of "mini-forums" would let you do it more often, with less
work on your part, and potentially get more exposure (to not just the faithful)
Q2: why would people using this software want to come to such an OGF forum when they
can just go to the software provider meetings (globusworld, condor week, etc.)
A2: because none of us paying people to build and operate grids want to invest in sending
them to five or six software-provider events every year, and having the providers (including
the ones who are not, but want to be, your provider) in one place facilitates speed-dating.
Q3: won't this make OGF appear to be pushing certain providers, potentially the wrong
ones?
Q3: yes, if we were talking about restricting providers. But we're not.
</Interanal Quote from another thread>
In talking to Ian Foster (Globus), Miron Livny (Condor), and Rich Wolski (UCSB- Network Weather Service and batch queue prediction tools) I have
already gotten their take on Q1. They all basically view this as a refreshing approach and one they would like to participate in.
As for Q2 - Speaking with my software consumer hat on, I already average 20 TeraGrid people at OGF and I would send 15-20 more if we had this kind of
forum (I can picture their faces; I am not talking in the abstract). [note- I am not viewing this as a scheme to get more attendees - the point I am
making is that OGF would be more valuable to me, a stakeholder]
CeC
|
