All, This one email contains the basic gist of the discussion thread from the end of last August. The upshot was to have some passionate individual(s) survey and analyze what is needed across these projects. Good idea but easier said than done! --Craig
Delivered-To: grdfm-gfsg-outgoing@mailbouncer.mcs.anl.gov X-Original-To: grdfm-gfsg@mailbouncer.mcs.anl.gov Delivered-To: grdfm-gfsg@mailbouncer.mcs.anl.gov From: "Alan Blatecky" <blatecky@unc.edu> To: "'Craig Lee'" <craig@rush.aero.org>, "'Catlett Charlie'" <catlett@mcs.anl.gov> Cc: "'Linesch, Mark'" <mark.linesch@hp.com>, <gfsg@ggf.org> Subject: RE: [gfsg] Information for next weeks GFSG policy call (resend) Date: Thu, 1 Sep 2005 07:59:30 -0400 X-Mailer: Microsoft Outlook, Build 10.0.6626 Importance: Normal X-Scanned-By: ITS Messaging Systems (outgoing) on 152.2.1.139 X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at mailbouncer.mcs.anl.gov Sender: owner-gfsg@ggf.org X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at mailbouncer.mcs.anl.gov X-MIME-Autoconverted: from quoted-printable to 8bit by rushe.aero.org id j81C1FI29142
Craig
Thanks for the clarification on what you meant by survey...it makes much more sense. I had to smile at your question about being "Boss of the World" as it is a good way to step outside of the problem and think of different approaches and solutions.
Your question of what can GGF do is right on target, especially if GGF is willing to be more flexible about its mission and what it is willing to focus on (and or can afford to do). Some of Geoffery Fox's comments and thoughts on communities come to mind.
Alan
-----Original Message----- From: owner-gfsg@ggf.org [mailto:owner-gfsg@ggf.org] On Behalf Of Craig Lee Sent: Thursday, September 01, 2005 7:39 AM To: Alan Blatecky; 'Catlett Charlie' Cc: 'Linesch, Mark'; gfsg@ggf.org Subject: RE: [gfsg] Information for next weeks GFSG policy call (resend)
Alan & Charlie,
By "survey" I did not mean creating a survey form and getting it filled out by epsilon number of people. I meant having some passionate individual(s) scrutinize all of these projects and the principle decision-makers involved, and make effective recommendations for convergence. Clearly this should be done sooner than later.
I'm sure there have been many, many meetings between these projects that have already failed to produce any kind of agreement on many issues. But is this a completely un-manageable process? Is there anything that can be done to effectively promote convergence? What can be done sooner than later?
If you were Boss of the World, what would that be?
Assuming there are hard problems that can't be solved by dictate, is there anything GGF could do to help avoid these problems going forward? Can we codify a strategy? Could we say anything more substantial than just "Don't change protocols every six months", or more concrete than just "Plan for change"?
Short of locking everybody in a room and not letting anybody out until agreements have been reached, what can we do? What kind of combined "summit meeting" could hopefully make progress? How can GGF go from generalities to specifics that are genuinely effective?
--Craig
At 12:56 PM 8/31/2005, Alan Blatecky wrote:
I have to second Charlie's comments. I also doubt that a survey will be able to accurately capture the "differences" between projects and/or areas of divergence, as the answers will depend on who filled out the survey (different participants in the same grid will have different answers).
Alan
-----Original Message----- From: Catlett Charlie [mailto:catlett@mcs.anl.gov] Sent: Wednesday, August 31, 2005 3:47 PM To: Craig Lee Cc: Catlett Charlie; Linesch, Mark; Alan Blatecky; gfsg@ggf.org; Gregory Newby Subject: Re: [gfsg] Information for next weeks GFSG policy call (resend)
Craig-
You are a good straight man- i.e. your response is a good illustration of the architect/builder disconnect I mentioned before.
My builders can't wait for a survey or a paper that realistically won't come out for a year. They are making decisions today with real operational systems and real users waiting for things.
As for meetings, I would bet there have been a dozen substantive meetings between peer grid projects just like the one Satoshi described between NAREGI and EGEE (I can tell you I've been involved in six of them!). That's crazy, and my suggestion is that those folks who are talking should combine the discussions and have FEWER meetings, not additional meetings.
CeC
On Aug 31, 2005, at 9:23 AM, Craig Lee wrote:
All,
Besides just organizing more meetings (and potentially just more hot air), what jumps out at me is the need for a survey project and perhaps a survey article that specifcally catalogues the design decisions made by all the major grid projects mentioned in this thread (and others), in each of the major technical areas, and what drove these decisions. Fox's comment that each project feels that it has captured the best possible software stack may, in fact, be true _in the context of each project's perceived goals and constraints_. Different projects may be driven to different conclusions by competing technical and project management issues and also _risk tolerance_. (As a simple example, the larger a project is, the more risk adverse it may become since having to upgrade/maintain software at all sites may become excessively difficult.)
We may all have a gut feeling for what is causing things to diverge (e.g., project schedule pressure -- I've gotta have something that works _now_ to support $100M of development in the application domain) but making a systematic study of the problem could help with concrete recommendations and courses of action to promote convergence. This could be the deliverable of a workshop series.
Naregi, EGEE, TeraGrid, OMII, EGEE, eScience, OSG, DataSynapse, United Devices, Platform, Sun
Technical areas: Security model Information model Job description model Workflow management model Auditing/Accounting
Another matrix...
--Craig
At 09:25 AM 8/30/2005, Linesch, Mark wrote: