Peter and I just met and discussed the disk issue. I temporarily reduced
the local user space (from 12TB to 3TB). In total, we have 24TB in
reserve (unmounted). I think we can hold 2x AODs and can support
simultaneous analysis and data transfer.
So far we only have one gridFTP server for data transfer at ~ 50MByte/s
(thus ~ 3 days in for 10TB in a good scenario). We are working with
Panda team to test our SRM installation, which can double the transfer
rate under existing configuration.
Can we meet this Wednesday at 9am ?
Wei Yang | [log in to unmask] | 650-926-3338(O)
Young, Charles C. wrote:
> Hi,
>
> Based on reports at today's US ATLAS FDR meeting (http://indico.cern.ch/conferenceDisplay.py?confId=27938), one complete copy occupies
>
> ~10 TB for AOD
> ~50 TB for ESD
>
> Tier 2's are expected to have all AOD. Tier 2's are not required to have ESD, but can subscribe to as much as they want.
>
> After the above datasets have been produced and transferred, we will stop *producing* more data -- it cannot produce fast enough to mimic the ATLAS detector anyway. However, we will continue to *transfer* the same data over and over for (I think) 3 more weeks. The goal is to demonstrate system stability. Another goal is for Tier 2's to see if they can support analysis on the first dataset while more data come in.
>
> Someone asked if we need more space for a second copy of AOD. This is a likely scenario in real life when we have updated constants, code fixes, and so on. There was no clear statement how much extra AOD space is needed. I suggest we budget for a full second copy, i.e. ~10 TB.
>
> More issues will no doubt crop up as we get going with FDR. Do we need to meet weekly for a while? Cheers.
>
> Charlie
> --
> Charles C. Young
> M.S. 43, Stanford Linear Accelerator Center
> P.O. Box 20450
> Stanford, CA 94309
> [log in to unmask]
> voice (650) 926 2669
> fax (650) 926 2923
> CERN GSM +41 76 487 2069
|