Testbed Support for GridPP member institutes
> [mailto:[log in to unmask]] On Behalf Of Simon George said:
> I also had the impression from the discussion with ATLAS a
> while ago that
> they were surprised that our SEs could not be regarded as
> permanent, high quality, backed up storage.
What do we actually mean by "backed up" here? Do most tier-2s really
just have "jbod" storage with no redundancy at all? If people have tens
of TB of disk I would expect some kind of RAID and hotswap spares, and I
would guess that's what atlas have in mind (plus some commitment not to
just delete files or remove SEs without warning).
> I think we do need ways to distinguish between the differing
> qualities of SEs.
We could try, but if you want to represent the different possibilities
it gets fairly complex, e.g mirrored RAID has rather different
properties to a secure room full of archive tapes in a different
building ... and the fact that failed PSUs may only be swapped once a
week is different again. And then you'd have to ask what you expect the
VOs to do with the information when they have it.
> I know for ATLAS, long term availability and uptime are important
> factors.
As commented yesterday, uptime is mandated anyway, you have 72 hours to
fix things and must have 95% availability per year!
Stephen
|