Hiro, I thought it might be useful to push up some concept visuals for autonomic/automated provisioning, that may assist in generating some thought and discussion at the face2face. Pete. -----Original Message----- From: owner-ogsa-wg@ggf.org [mailto:owner-ogsa-wg@ggf.org] On Behalf Of Hiro Kishimoto Sent: Wednesday, August 03, 2005 7:35 PM To: Milojicic, Dejan S Cc: cddlm-wg@ggf.org; ogsa-wg@ggf.org Subject: [ogsa-wg] Re: [cddlm] Notes from the OGSA EMS & CDDLM meeting Thanks Dejan for the meeting notes, I've put proposed joint session agenda into F2F schedule and update attendees list accordingly. https://forge.gridforum.org/projects/ogsa-wg/document/OGSA-2005-Aug-F2F- agenda If you have any farther questions/comments, please let me know. ---- Hiro Kishimoto Milojicic, Dejan S wrote:
By phone, 8/3, 2pm-3pm PST.
Attendees: Mark Morgan, Hiro Kishimoto, Chris Smith, Stuart Schaefer, Jun Tatemura, Ravi Subramanian, Mike Behrens, Steve Loughran, and Dejan Milojicic.
Notes taker: Dejan Milojicic.
Agenda:
1. Go over the two documents (CDDLM Foundation document for EMS people
an the OGSA architecture document for the CDDLM people.
Dejan suggested that CDDLM is part of the EMS, but there are other pieces which are out of scope.
Chris asked about activity v. solution. The answer was that we support
solutions not activities.
Hiro: can we also support C code, Cobol, etc. The answer is yes, but we need to support the environments.
Ravi: we need to resolve the terms between the two groups prior to the
meeting.
Hiro: there is a glossary document. Good suggestion (Dejan), that way we can get consistent independently. It is linked off of https://forge.gridforum.org/projects/ogsa-wg/docman/draft-ggf-ogsa-spe c/ en/23.
Ravi: There is a space for provisioning in the EMS document, we try to
plug in the CDDLM as a virtual exercise. We try to effectively see if CDDLM is compatible with OGSA. Chris, there are issues with containers.
Chris: extending JSDL (extensions) with CDDLM or embedding the schemas. Steve: it is batch -centric. Chris: it is not really. Ravi: keep provisioning separate from JSDL. Dejan: summary, this is a good proposal, but for simplicity reasons, we'll keep it out of our immediate objectives for the time being.
2. Agenda for F2F (8/16, 1:30-5 in Sunnyvale):
Action preparation for the meeting: sync up with glossary (Ravi: also check up the glossary if we have any comments).
A) (1:30-1:45) Introduction (learn more about EMS broader picture, in the past wee overviewed CDDLM) B) (1:45-3:15) Defining architectural relationship (This is really related to the pictures we discussed above, Fig 5 in the OGSA arch & Fig 7 in CDDLM Foundation) C) (3:15-3:45) Indebt overview of the CDDLM interfaces (if needed) D) (3:45-4:15) Description and integration of ACS & CDDLM (Ravi: added
BES as well). E) (4:15-4:45) Integration roadmap F) (4:45-5:00) Next steps
(Dejan: above times are approximate, they may change as needed, e.g. if we cover some of the later items in earlier sessions)
Ravi: is this similar to BES interfaces activities. Chris: this was exactly how I was thinking wrt BES container. We can jump ahead to conclusion that CDDLM also created container. it is definitely in the context of BES cause we will run through the BES and we will automatically se how it fits with BES. The same applies to RSF. It is almost looking through a use case how do we do with something that is given to job provisioning.
Hiro: There is one more important component, ACS. Dejan/Steve, we came
to some agreement with ACS at GGF 14 and there is no need for changes to CDDLM wrt ACS. Hiro: ACS had a F2F meeting in North Carolina. Mike prepared a document describing relationship between CDDLM and ACS. Ravi: we can add the ACS diagram to the figure. (as well as BES), this way we can tackle multiple birds at the same time. Mike: ACS has taken requirements wrt contents, there are some security considerations and creation time issues that would require something other than URL, maybe service invocation, we can go over these details at F2F.
3. Techniques and expectations:
Dejan: informal approach, discussions, drawing pictures, agreeing on interfaces, it is ok to have a few more open issues, but resolve them at latest at GGF15.
Ravi: can someone bring digital camera, take a snapshot and then send it, put it on Webex.
Hiro: expects architectural diagram and text describing components, so
that they can come into the next version of architectural document.
Ravi: when is the time (Hiro: 1:30-5pm)