Participants

• Sergio Andreozzi (CNAF)
• Riccardo Zappi (CNAF)
• Laurence Field (CERN)
• Felix Ehm (CERN)
• Stephen Burke (RAL)
• Paul Millar (DESY)
• Maarten Litmaath (CERN)
• Shiraz Memon (FZJ)

Minutes

• SA: StorageSharePolicy might not be right place for transport protocols; if you mean endpoint transfer, then why not use StorageEndpoint with different type?

• SB: do we need to suppor Classic SE?
• SB: do we need transport/access protocol related to shares or are they global for the SS?

• ML: we probably need to publish that unsecure rfio or dcap are available only locally (as a flag in the endpoint?)

• SB: how do we clarify which computing resources can really use the locally-available rfio/dcap?
  • we need something similar CESEBind

• ML: what about if rfio can be used from a UI? the CESEBind won't work

• ML: Jens in the past mentioned use case of hw-depended available protocol
  • might be overkilling use case, mentioned for completeness

• PM: why do we need to publish Storage Components?
• SB: in GLUE 1.3 it was left out because we've not found use cases

• SM: Transport Protocol type should be an enumeration

Changes:

• for the time being, we agree to remove storageResource/Component
• move attributes of StorResource in Stor.Service
• multiplicity checks
  • StorageEndpoint (0..*) --- (1 AGGR) Stor.Service
    • agree to keep this way
  • StorageShare (1..*) --- (1 AGGR) Stor.Service
  • StorageEndpoint (0..*) --- (0..*) Stor.Share

• Next telecon: Friday 25 jan, 4 PM Rome