Burke, S (Stephen) wrote:
>LHC Computer Grid - Rollout
>
>
>>[mailto:[log in to unmask]] On Behalf Of Ben Waugh said:
>>I'm giving a talk about LCG tomorrow and wanted to find an up-to-date
>>number for the total data storage available on LCG. The gstat
>>page gives
>>the answer as 97 PB, but 93 PB of that is at IN2P3-CC. Is
>>this a mistake,
>>or does IN2P3 really have over 20 times the capacity of the
>>rest of LCG?
>>Does anyone have a realiable figure?
>>
>>
>
>The amount of tape storage is pretty much undefined, maybe they just put
>the maximum capacity of the building! Probably the best you could do is
>count only the disk storage, and then mention that there are tape stores
>as well. In theory the type of storage is in the GlueSLArchitectureType
>attribute in the info system, but (as usual) it looks like not everyone
>is setting it correctly:
>
It should be notice that we can't publish anything else than disk as
SEType for our SEs (even if it is a tape system SE) due to a bad
behaviour of the lcg-cr command.
Still, we have to publish a distinction between tape system SEs and
disks SEs, so we publish "mms" for GlueSLArchitectureType vs "disk".
There is no clear instruction about what to publish for these values,
contribuing to the global confusion on this matter.
I allready talked about this problem with Laurence, Piotr and James
Casey and made some suggestions, I look forward for the next decisions
about this.
Jonathan
|