OGSA Teleconference - 28 February 2005 ====================================== * Participants Jay Unger (IBM) Jem Treadwell (HP) Ravi Subramaniam (Intel) Latha Srinivasan (HP) Andreas Savva (Fujitsu) Steven Newhouse (OMII) Mark Morgan (UVa) Tom Maguire (IBM) Fred Maciel (Hitachi) Hiro Kishimoto (Fujitsu) Andrew Grimshaw (UVa) Dave Berry (NeSC) Mike Behrens (R2AD, LLC) Minutes: Andreas Savva * GGF13 update - Hiro will add the dial-in number to the GGF13 sheet - ByteIO BoF is at the same time as the Basic Profile review (session #2 conflicts #11) - BES BoF conflicts with Data architecture (session #8 conflicts with #12) - Andrew is therefore double booked; he will probably be in BES. - #3: Probably do this as a BoF as agreed at the February F2F. The proposed group is probably in the Architecture Area so Andrew is the AD. - Andrew to arrange logistics - Also contact Manuel to find out if he'll be attending GGF13 (Confirmed attendance during the call) - #4: No draft document. Not sure how advanced things will be but will prepare for a discussion. Maybe an open-floor session without long presentations - #5: No draft document. - #6: Basic profile: Was naming and security pulled out? If so how far does this session cover. - Naming: probably everything relating to ws-addressing goes to another (naming) profile. (Something to discuss later.) - Security: not sure whether taking it out completely is good or not. Security is something to 'bake-in' to every spec. Tom and Dave decided to keep the current text in the BP and revisit this issue later. - #9: - What is the intended outcome of this session? - It's not clear whether this is a discussion/educational session or not, or how things will fit together. More likely a discussion, exploration style session. - Is there a list of standards? Or is there any consensus on what the architecture needs? (model driven) - OGSA version 1.0 has a resource management section but the ideas have not been applied consistently throughout. (Which entities are resources and which are not. What does that mean in each case, and is the treatment consistent and so on.) - Part of the work is in the EM/data design teams; with some/other work in the RM team. - Fred is working to bring things together for this session - (Need traceability of requirements back to OGSA version 1.) - Gaps in profiles and how to address them - A side-discussion from the EM profile discussion. - What to do with such gaps? - If something is really missing and it is needed - Ask the spec owners to fill it in (or find owners) - Fill it in the profile (assuming a relatively small and easily bounded gap) - Deliverables review - Three docs submitted so far - Two more from the original list remain to be submited: Data and Roadmap - Roadmap - Ian is working on the Roadmap. Hiro will check the status with him. - Agreed that the draft should go through a group review first before sending it to Stacey. - Data - Ask DaveB. (Hiro) - Visa for Korea: - [If you are unsure contact the nearest Korean embassy.] - A tourist visa should not be used if you are travelling for business. - Information on how strict Korean Immigration is varies. - Jay has more details from the Korean embassy. Email him if you need it. * EM refactoring doc (20050228) [Posted to gridforge after the call https://forge.gridforum.org/projects/ogsa-wg/document/ogsa-ems/en/2] - Fig. 2 : business process/ workflow / ... / tasklet - Trying to describe the scope of area (so includes business process to tasklet) - Including business process here might raise concerns since there is a lot of work outside the grid on this. - Clarified that the intention is not the modelling of the business process but what is needed in the environemnt to support its execution. - Terminology issue: Should these terms be used (clearly overloaded) or should different terms be used (ref: Terms proposed by Andrew & Mark) - Fig 3 (demand - supply) - Added "execution plane" and "control plane" - Control plane 'controls' execution/provisioning activity - Trying to draw out similarity with other capabilities, e.g., data - Resource model would be part of a Resource Management framework - The group is working from both ends but there is a large gap between the lower level work (leading to a profile) and work at this level. - Obviously the situation is not ideal. It is a balance between the desire to do something that will last vs not being able to achieve anything within this year (and taking on the risk of some re-work later on) - A concern raised is whether there is really an incremental path to fill the space or not. - BES focus is 'job focus' ('work push'). Ravi also wants to have the opposite perspective of a resource 'pulling' a job ('work pull') included. - Could be done as extension to the current proposal at a later point (in any case it is not precluded) - OGSA EM v1 identified some core functionality; BES focuses on the simplest function within that picture. The expectation is that there will be a build-up from that base to do other functionality at a later stage. (E.g., CGS) - Next steps? - CGS is one likely candidate - To try and define CGS the following would also be needed: - How jobs describe resources (e.g., JSDL) - How resources describe themselves - And also have to 'unify' across capabilities (Ravi) - Jay argued that this is 'not a unification' but a 'relation'. - Model of how to manage data as data (can be used by the manager) - And a view/facet of that model that can be projected or used by the CGS - This view does not need to be the same as the manager's; only what is required by the CGS - Agreed that this is a good topic to discuss at the EM architecture session. Ravi to prepare for kicking off this discussion. * BES BoF preparation - Andrew, Mark, Steve have had a short call on logistics (who will present, etc) - Andrew to send out draft answers to the 7 questions - Steven will send out revised version of the draft API document (but refinement of this document is not the main focus of activities at the moment) - Will review these again before GGF13 (probably Mon Mar 7) * AOB ** Resource Management followup - Fred is working on a concrete proposal on how to proceed: - Resource modelling has to be done by both sides at the same time. The two sides being the capability (e.g., EM, Data) and the resource modelers (e.g., straight with the DMTF via the resource management team) - A draft document is nearly ready and will be distributed soon. - The CGS co-chair, Tom Rooney, is also interested and is getting involved in this activity. ** BP review postponed for next call