Phone Meeting

• Date: 3 May 2007
• Time: 4 PM CEST (conversion to other timezones)
• Calling number:
  • via skype (free): +99008275446903
  • via traditional phone using these numbers and this conference room number: 5446903
    • not free, but cheap

Agenda

1. preparing for OGF20 participation
   1. Joint Session
      1. template for group presentation
      2. gather main questions to be answered during the meeting
   2. GLUE Use Cases session
      1. review the the current user stories and verify if further contribution will be delivered
   3. GLUE Entities
      1. present the initial sketch on entities
2. AOB

Minutes

• Participants:
  • SA: Sergio Andreozzi (INFN)
  • ML: Maarten Litmaath (CERN)
  • SF: Steve Fisher (RAL)
  • SB: Stephen Burke (RAL)
  • JP: J.P. Navarro (TERAGRID)
  • MV: Matt Viljoen (RAL)

SA: comments on the adopted conference system: it worked really well for someone, only bad sound for people calling from UK using fixed phone; after 2/3 trials, they got good service; to decide if to stick to this system or ask Laurence for using CERN conf system for future conferences (all people would need to call to Switzerland)

SA: agenda review;

SA: present goal and scope of joint session; ask to SF if it is meaningful to have also a presentation from SAGA WG; SF suggests to check with Andre Merzky (unfortunately a SAGA WG session is in parallel to the joint session)

SA: present goal and scope of GLUE Use Cases session; describes why user stories were selected for gathering requirements and testable solutions

JP: asks if there is any constraints on user stories categories

SA: we should mainly concentrate on real experience; user stories should mainly focus on what current schemas cover; nevertheless it is suggested that new user stories should be anyway collected and marked as new; the importance to meet the deadline of GLUE 2.0 definition in public comment by the end of the year; if there is room to discuss new user stories, no problem

SA: asks if people have experience with user stories writing; not relevant experience was reported

SA: presents goal and scope of GLUE Entities session; describes the new concepts related to the computing element with focus on the important introduction of the share concept and the independence of how actually shares are configured (this is one of the week points of GLUE 1.x); an attempt to differentiate computing services vs. computing resources is also provided;

SF: do we really need to differentiate them?

SB: there are use cases for which we need; e.g.; if we want to know how many resources are actually available in a Grid (avoiding double-counting)

JP: about execution environment, there is a way to model basic execution environments with selectable components from a menu? in teragrid this is a real use case that is going to be published with next Globus release

SA: an approach could be to model the basic execution environment by using the execution environment entity; the options are provided via the application environment concept; anyway I suggest that JP writes a user story on this

JP: ok

ML: (1) it is nice the new concept of share, but we should carefully evaluate how it impacts on current middleware; e.g. job submission is based on placing jobs to queues; (2) we should try to get right the sharing of the same resource among many services

SA: about (1), I see the queue as a potential attribute for a share; the queue is not part anymore of the identification of an entity; just an attribute that can be used at submission time; (2) yes, our experience + user stories + acceptance tests should help on this

discussion on the authorization entity; SA and ML explain to JP how group-based allocation works in gLite; JP explains the approach in TERAGRID; they are different and we can learn from each other

ML: there is plan to deploy GLUE 2.0 schema also on pre-OGF services? (e.g., in EGEE for current info service and using current computing element interfaces?)

SA: there was no talk about this, but we should; we cannot expect an instant transition; GLUE 1.3 info and GLUE 2.0 info should coexist for a while possibly using the same info service instances; to be put as a requirements

SA: asks to SF if he can contribute a service definition compliant with SAGA view; he will provide that

SA: informs that OMII-Europe project will allocate a budget to INFN in order to work on info modeling for computing resources and actual implementation for OGSA-BES in gLite and UNICORE; this means two persons for one year starting probably from 1 June 2007; important contribution to speed up the activity

Actions:

1. SA: send email to Andre Merzky to check if a SAGA WG representative can contribute SAGA viewpoint
2. JP: to write one/many user stories about the dynamically configurable execution environment
3. SF: provide draft description of Service matching SAGA view