WLCG Information System Evolution Task Force

513/R-068 (CERN)



Show room on map
Meeting to discuss the evolution of the WLCG Information System


Alessandro P., Andrew ,George , Florin , Oxana, Jarka, Rob, Giuseppe, Stephan, Marina, Scott, Alessandro G., Julia, Salvatore, Maarten

Policy for introducing new service types

Julia suggested to create a mailing list with representatives from OSG, EGI and WLCG IS Task Force.

Introducing a new service type should require approval of members of this mailing list who agree on the name for the new service type which will be then consistently used by all systems like OIM, GocDB, SAM, experiment -specific systems, etc...

People agreed with this proposal. Julia will create a mailing list with members communicated to her by the EGI and OSG representatives. The new policy will be announced at the next WLCG operations coordination meeting (2nd of March).


Storage service description

Discussion after Andrew's presentation

For complete description of the storage services LHCb relies on Dirac. Though most of this information is also provided via BDII. How synchronization between the two is ensured? When  any relevant change is performed in BDII, there is a notification via Dirac client. The update of information in Dirac DB is manual. There are only 20 sites with storage, so it is not a big deal.

Whether current information in GocDB is enough to  handle downtime ? Yes, it is quite sufficient.

Alessandro worked through the attached schema. Further discussion

Should be there a reference between the Data Store and Data Share? Not necessary. Data Store could be an optional object, not strictly required.

What do experiments rather than ATLAS think ?

Andrew for LHCb: Looks good, should be able to describe storage service for LHCb.

Maarten for ALICE : In principle ALICE is planning to rely on its own topology and configuration system. The structure shown by Alessandro looks consistent with GLUE2 which took into account all experience accumulated over last years, therefore, there is no worry that the schema won't be able to describe what is required by the experiments

Stephan and Giuseppe for CMS: CMS would need more time to comment on storage description. Currently CMS priority is description of the computing parts rather than storage. This work is ongoing with the CRIC team.

Scott for OIM: Are any changes in OIM would be required to be compatible with this schema?

No, there is no requirements for OIM or GocDB, the proposed structure is for CRIC.

Andrew: Whether DPM and dCache are involved in the discussion?

Not yet. There was a storage accounting discussion during the pre-GDB which is certainly related to the storage object structure, but there was no discussion regarding storage object structure as such. Should discuss it with all storage providers. There will be a data management steering group meeting next Monday, Julia will ask to organize a dedicated discussion.

Another aspect is how this information will be filled  into central repository (automatically vs manually). Alessandro thinks that there should be various options. We should not wait for all information providers being compatible with the storage structure. This is static info, not being frequently modified.










There are minutes attached to this event. Show them.