Hi Tim,
On Sat, May 07, 2005 at 02:09:22PM +0200, Peter Elmer wrote:
> On Sat, May 07, 2005 at 01:02:42AM +0100, Adye, TJ (Tim) wrote:
> > The first is best for slac and ccin2p3 which don't use BbkImport and
> > probably have all components. The second is best for the rest, since it
> > only counts files that were actually imported, not whole collections.
> >
> > > SLAC 316TB (--dbname=bbkr14 and --dbname=bbkr18)
> > > In2p3 232TB (--site=ccin2p3 --dbuser=bbranyuser)
^^^^^
BTW, I suspect there is some double counting going on here from the
bookkeeping, since I see now that Jean-Yves wrote that they have imported
~115TB to Lyon.
Pete
> > > RAL 65TB (--site=ral)
> > > CNAF 50TB (--site=cnaf)
> > > FZK 26TB (--site=gridka)
> > > Padova 34GB (--site=padova --dbname=bbkr14)
> <...>
> > Was that what you were interested in?
>
> That is actually also very interesting. (I see Jean-Yves is slowly pulling
> in everything!) I was thinking earlier of only the disk space and # of
> servers, but perhaps this is the table to fill in:
>
> data
> total disk servers in
> data cache disk cache #client cpus
> ------- ------- ---------- ------------
> SLAC 316TB 160TB(*) 30(*) ~2000(**)
> In2p3 232TB 42TB 15
> RAL 65TB
> CNAF 50TB
> FZK 26TB
> Padova 34GB
>
> (*) This was _supposed_ to go up since there are another 10 "kan" servers.
> What happened to that? (It also doesn't include the skim output system,
> which is 11 servers + some smaller amount of space.)
>
> (**) I've seen up to 2200 client cpus in the past. I just looked for the
> past week and see ~1500 were in use. (What are the rest at SLAC doing?
> There should be 3200+ cpus total between barbs, nomas, toris and dons,
> plus the broncos, although I guess the broncos and barbs are now being
> retired.)
>
> Poor Padova, only 34GB of micro/mini data... ;-) Well, they compensate
> because they have 250+ TB of raw data in xtc file format... Roberto Stroili
> was asking last week if we can setup reading of xtc files via xrootd for
> BaBar, too. At one point Alvise put something together to do this with the
> original XTNetFile client implementation, but now that we have the posix
> client we should just do this properly for BaBar.
>
> Pete
-------------------------------------------------------------------------
Peter Elmer E-mail: [log in to unmask] Phone: +41 (22) 767-4644
Address: CERN Division PPE, Bat. 32 2C-14, CH-1211 Geneva 23, Switzerland
-------------------------------------------------------------------------
