HEP Tape Archive definition
Proposal
The working group should deliver a proposal for use in upcoming DAM discussions which clearly states the view from the tape experts regarding the use of tape in the future.
The proposal should address all issues regarding the use of tape as ‘true archive’
- Define the tape experts definition of ‘archive’
- To support the DAM discussion started at the Jamboree
- To prevent the use of tape as random access device in the future
What makes archives true (German you had more items for this list. Can you add them please?)
-
- German's list:
- Data needs to be organised in Containers (example: tar files) which are opaque (the container is the storage unit, rather than the individual files).
- Containers need to have a minimum size, in terms of volume, files, or even age.
- Containers should become read-only once they are migrated to tape. This is to avoid fragmentation and versioning within container payloads.
- All required metadata for the payload should be within the container itself. The metadata does not need to be seen from outside the container.
- End-user access to containers should be controlled or even better, restricted (ideally: banned). Alpha users from experiments are responsible for container get/put operations.
- A container should have a maximum lifetime defined (infinite is possible, but needs to be set explicitly).
- Data 10 years life time? more?
- Near-line or shelf?
- Do we need duplicates
- Lifetime on data needed?
- Simple IO: put, get , delete (do we want or need ‘ls’?)
- Metadata handling
- Type of protocol to access the archive
- Restrict access to the archive to prevent misuse and propose how to limit this access
--
VladimirBahyl - 07-Jul-2010