See this note archived at:
the Application Workshop discussed some of the action items from the Madison US LHC OSG Technology Roadmap meeting. The outcome of the discussions on metrics and goals for OSG 0.2 and the planning and definition of OSG Release 0.4 are provided as input to the Technical Groups and Activities, with a goal to have them submitted for endorsement by the council before the July consortium meeting.
The consensus at the meeting was to propose a Blueprint meeting the 29th and 30th August. The goal of the meeting will be to deliver a Requirements Document for Accounting, Monitoring and Information Services for the OSG Technology Roadmap.
I thought the Applications meeting was useful and hope that the discussions on collaboration and action items still look good after a weekend. We had input from more than 10 VOs. People who did not attend are invited to read the posted talks and ask questions on the osg-slac0605 list.
US LHC OSG Baseline Services roadmap document is
Rob and Frank will be posting a new version soon.
Define Metrics for OSG 0.2:
Efficiency of Grid Infrastructure and Sites (ie exclude efficiency of applications)
Total amount of data produced
Total amount of data ingested by applications
% of available CPU consumed – not a metric as impossible to measure denominator. Sample?
Total CPU consumed.
Number of science papers including results from processing done on OSG
Number of different applications run on OSG
Ratio of number of Trouble Tickets to number of ?
Average Time to closeout a Trouble Ticket.
Average Time to bring up a site.
Application Administration FTE load
# of different users submitting jobs.
Fraction of resource usage that is Opportunistic
Definition of OSG 0.4 Release. Proposed for circa November 2005. Additional services and functionality from OSG 0.2:
Edge Services Framework
GT4 Gram as default job dispatching interface
Policy for quota’d space per VO (50G each)
Support for Late binding WMS
Workload Management System
Support for Late binding
Job inspection service
Guaranteed job exit codes – batch job executor sends exit code through xxx
Extensible information about service and resource
SE based on SRM
Storage role based access control
resource information repository, interoperable with LCG and partner Grids and with VO specific monitoring infrastructure.
Metrics instrumentation and presentation
Federation infrastructure with partner grids.
FQAN name space policy
Redundancy and fault tolerance (e.g. for GUMS and VOMS implementations)
Support for multiple implementations (ie not just INFN VOMS)