* Storage Classes use cases. Use of different disk pools for transfer
and anaylsys (9,10).

Artem: It's not enough to specify tape1disk1, since experiments want to
keep files on the pools where jobs access them, not in transfer pools.

Graeme: experiments need to keep some files on transfer pools too.

Jos: so far had special arrangements for transfer/analysis pools for each
experiment.

Graeme: can we use storageSystemInfo of to provide hints?


* keeping disk only data on tape pools? (11)

Artem: if a site needs to host the data for which it's not a custodian
(Atlas aod replica), will it be more usefull to keep such data on tape
anyway?

Jan: this is what CERN is going to do.


* Missing,lost files(14). Discussion regarding updating site's MSS catalog,
LFC:

Graeme: site's can't update catalogs LFC, as it's secure.

Jan: so far working individually per user,experiment. Need more scalable
way.

* srmChangeForSpace (17,18): should tape0->tape1 be supported too?

Mark: whether tape0->tape1 is supported depends on particular
implementation. Should not matter from operational point of view.

Jan: for Castor doesn't matter.
* srmBringOnline(19) Dividing pools into analysis, transfer

Jan: Already doing this at CERN. But all pools are accessible with
gridftp too.


* storing raw, aod data separately? (21)

Jan: It's in experiment's best interest.

Graeme: use attributes with hints in SRM calls?


* Further plans for the WG activity

Mark: would like to invite dCache developers.

Jos: would like to invite CMS person to describe howCMS is going to store
data, use SRM etc.

Decided: to invite CMS representative.

Next meeting: the week of Jul 24-28 or Jul 31-Aug4, TBD