First thoughts:
Overall goal: give users guidelines what to do in order to exploit
multiple infrastructures
* Negotiation of usage
- Is there a notion of a "home" grid?
this would allow to reuse techniques used in mobile telephony but
how do "multi-Grid" VOs relate to that?
- Access Policies
Supercomputer Grids have tighter regulations for access
does this call for different VOs or could existing mechanisms
be exploited to enforce these policies?
do people have to sign with each resource center separately?
can we come up with a common user registration? Can we have a
superset of
information sufficient to allow access to all participating grids
or would a common denominator do?
- CPU and data usage
who are the actors: VO, user? Grid, resource provider?
Could there be a common application form?
different models exist:
* guaranteed resources
* opportunistic resources
have allocations on guaranteed resources be negotiated with each
Grid individually?
how can opportunistic resources be negotiated?
how can data be exchanged and managed between infrastructures
- network connectivity
does the existing/planned network have enough capacity/right
structure
when the federated Grid becomes reality?
research network vs. private networks
* Accounting
- tied into "home Grid" discussion
- is there a need of a common way how Grids account with their VOs
and who would be the actors
- should there be a "bank" i.e. somebody who keeps the accounts?
- would you participate in the design of a standard accounting
query?
- would you push out account status to a "home Grid" or "home VO"?