Minutes attached. https://forge.gridforum.org/projects/ogsa-wg/document/minutes-20051130/en/1 -- Andreas Savva Fujitsu Laboratories Ltd OGSA Teleconference - 30 November 2005 ====================================== * Participants Hiro Kishimoto (Fujitsu) Fred Maciel (Hitachi) Tom Maguire (EMC) Steve McGough (IC) Mark Morgan (UVa) Takuya Mori (NEC) Andreas Savva (Fujitsu) Ellen Stokes (IBM) Ravi Subramaniam (Intel) Jem Treadwell (HP) Apologies: Steven Newhouse Minutes: Andreas Savva * November 28 minutes approved with no changes * UML tool update - RSM was chosen; no objections so far. - Ellen is following up on how to get licenses. - (It is used by DMTF already so it is possible.) * Meeting planning ** GGF16 - Proposal to organize just the default 2 sessions. - It worked well last time and also allowed WG members to attend other sessions as well. The separate one day F2F should allow enough time for more detailed discussions. - One session could be 'Outreach'. But 'Outreach' last time was not well attended. There are two possibilities: - Andrew G's proposal to have a 'Town Hall' style meeting instead. Talk with Andrew to get more details - An outreach session as a 'deep-dive' for a specific capability to introduce the group's work in more detail. - The second session was a joint-wg discussion. Last time it was EMS-CDDLM. Having a similar joint discussion with another group might be a good idea. - The deadline for session requests is Dec. 14. - Groups requesting sessions after Dec 2 must have an up-to-date charter; not a problem since OGSA's new charter was approved recently. ** GGF16 B2B F2F - OGSA-WG has been asked not to overlap its B2B F2F with GGF sessions. So this time the back-to-back meeting will be on Friday only. - A room at IBM Athens has been reserved. But it may not have internet access. - Hiro is also looking for a room at a University in Athens and if it offers internet access the meeting may be moved there instead. - In any case a final decision for the venue should be made by the end of December. ** GGF16/F2F Actions - Hiro will send out a topic request to the list - For both the GGF16 joint session and for the one day F2F. The choice for the joint session to be based on people's availability (and whether they can attend the Friday F2F) and how big the topic is (fits in 90 minutes). - Review choices next week ** Jan. 2006 interim meeting - Agenda reviewed once on a call. Hiro is updating it and will review it once more. - Hiro will also do a zoomerang survey to confirm people's availability. * BES Information model review - Ellen sent out a ppt before the call. Only slide 4 was reviewed on the call. (Slides 1-3 are carried over from an earlier presentation and slide 5 is redundant and will be deleted.) - Slide 4 is based on work between Ellen, Andrew and Mark and is a first step in working out the minimal set of required and optional information for a BES container. - At this point, this work does not consider activities that may be started within the container. ** Operating System: - It seems to be accurate for a server box but does it cover other possibilities such as 'Java VM' or batch system (e.g., PBS)? - This is probably covered by the value of OSType. - Is this the best way to map to CIM? - There are probably some missing bits. The motivation is for this to describe an operating environment - The attributes are constrained to be specific to a BES container but it is expected that there will be a new class in CIM (Service Container) and that the BES container will be the constrained profile version of this Service Container (and not a subclass). - The classes are not shown yet - Other attributes: These are attributes that need to be explored further. E.g., libraries/packages are actually part of the application model in CIM and it is not clear how best to fit them in yet. ** Processor: - What is the difference between the '# of processes' here and the 'number of processes' in the Operating System? - It is not clear and needs to be reviewed again. - 'Load for computer systems', '# of processes': They don't seem so specific to Processor. - (Have to also think of virtualization: What is real or virtual in CIM. This is an ongoing discussion within CIM.) ** Networking - This is just an initial attempt. Current scheduling systems do not require this kind of information. - Still it seems strange that even simple descriptions such as the number of cards or their bandwidth are not part of the Optional list. ** Disk - Not discussed yet. To revisit later. ** Filesystem - Why is 'casesensitive' or 'filesystemtype' optional? - Most systems do not use these for scheduling ** ServiceContainer - The attributes listed after the first two items in the list should be optional not required. ** Management interfaces - Which of these are functional vs management interfaces? There are different perspectives, with the consensus that 'Stop' and 'Start' are probably management but less sure about the rest. - This is another item to discuss again in more detail within the design team. ** Overall - Consensus that this is a good step forward. It does need more work. - How many pages would a completed profile take? - It should not exceed 50 pages. - Next review? - Ellen's schedule is full in December. Work will continue within the smaller group and it may be possible to have another review in January before the F2F. - Jan 9 is probably the only full OGSA call possible in January before the F2F(+). But a smaller group could use one of the call slots to have a review. (+) Note: GFSG meeting is planned for the week before the OGSA F2F. * MDC presentation review - Reviewed presentation posted by Ravi to gridforge. Fred made revisions on the presentation during the call. Summary of main changes below: - Add EMC to Contributors slide - 'Concepts and Fundamental' document should be changed to Architecture document - Add that OGSA profiles are styled after WS-I profiles and are extensions of WS-I BP etc - Naming - details are under discussion but at an abstract level the content is ok - Need more graphics for OGSA Data: Ravi to ask Dave B - Newer version for 'Typical Pattern': Andreas to send to Ravi (Done) - Add CMM GFD # and also do the same for other published documents showing on the slides. - Comments to Ravi by Monday latest. The presentation is planned for Tuesday. - Fred will upoad as next version, tidy up the folder and send email how to uploade new versions of existing documents. (Done) * Next call ** Monday - 'BSP - Secure channel profile' (draft expected Friday) - Architecture and Glossary 1.5 review ** Wednesday - Naming & EMS