Guys, this looks an encouraging start. Some of the questions raised have
already been recognised as 'known unknowns' in LCG and EGEE and I'll push to
get answers, plans, or decisions.
One example is the criterion for joining LCG (and later EGEE). This is
undecided but is actually orthogonal to our aim. We wat to put together a
GridPP Grid. You can all join that independent of your size. We may make UK
restrictions but we needn't be bound by LCG or EGEE decisions. LCG2 software
is the common factor between grids just now. All our UK sites don't need to
be in LCG but if they run the same software we have a flexible grid and some
sites can be both LCG and EGEE and GridPP sites. That said, the LCG test
zone is open to everyone just now so there is nothing to stop us all joining
so go for it.
A couple more points below.
> -----Original Message-----
> From: Testbed Support for GridPP member institutes
> [mailto:[log in to unmask]]On Behalf Of Steve Traylen
> Sent: 15 April 2004 17:24
> To: [log in to unmask]
> Subject: Minutes GridPP Testbed/Deployment Meeting
> Please send me privetly any corrections in particular
> if you were there and I missed your name apologies.
> Minutes GridPP Testbed/Deployment Meeting
> Thursday Apr 15th 2004 14:00
> + Question from Rhys. Where should I looks to see overall
> architecture document?
> Lots of blank faces but suggestions of EDG architecture
> documents and
> possibly EDG papers in the future. (Perhaps someone could
> provide links)
I'll try and find something. New EGEE sites have been asking the same
> + Matt, Which testbed for HEP production work?
> LCG2 is the definitive answer and to some point at least LCG2
> is the same as EGEE production anyway but LCG2 is what
> exists today and
> has the longest future.
Agreed but the decision is actually down to the experiments. They will
manage their own lists of sites to be used. Just advertising yourself as
supporting the atlas VO isn't enough.
> + Santanu. A question about different hardware within the
> batch workers and how
> this should be represented in information system.
> .. You can't , GLUE assumes all hardware is the same.
> Select some middle ground perhaps with the exception of
> memory where you should publish the smallest memory available.
GLUE has subclusters which should be heterogeneous. A cluster can then
consist of a number of clusters. That was the plan anyway. Whether CEs
actually map onto clusters I cannot confirm.