Speaker
Brian Davies
(STFC (RAL) GB)
Description
The Rutherford Appleton Laboratory (RAL) operates the UK WLCG Tier1 facility on behalf of GridPP. Tier 1's provide persistent archival storage (on tape at RAL) and
online storage for fast access data analysis. RAL is one of the few Tier-1s which
supports data management for all the major LHC experiments, as well as a number of
smaller Virtual Organisations. This allows us to compare usage of the infrastructure
across many different experiments and use cases.
In this post, we look at the distribution of file sizes between VOs and across
different storage work flows within the same VO. We also report on how this
distribution impacts standard operational procedures such as tape usage and
decommissioning of storage nodes.
Further, we also show how a hypothetical change on data storage policy of log files
would change the profile of filesize distribution of a significant subset of data
currently being stored at RAL and how this change would impact operations to the
advantage of the user communities.
Primary author
Brian Davies
(STFC (RAL) GB)
Co-author
Shaun de Witt
(STFC)
