Hi
There was an action / task in the group to look at the access patterns of data. Not meaning the local access of an individual file, which we have talked about here before, but the access of datasets across the UK.
ATLAS have a newish tool to help with this - Popularity: http://popularity.cern.ch
I am scraping some of the info of this page for the UK specifically,
http://www2.ph.ed.ac.uk/~wbhimji/Popularity/
If you hit dataset popularity table, then you can see the most popular datasets in the UK in the last 14 days and where they are.
You can also filter by a site (or Tier 2), (you may want to put a number in min requests < 5000 to see everything but > 1 so as to exclude a lot of junk) - it is a bit slow for this so you may just want to use the original popularity page.
For fun, I put a "Site Usage" button to see which sites are getting the most action but there are too many pieces in this pie.
We can build on this, if we want (and I am republishing this info as JSON http://www2.ph.ed.ac.uk/~wbhimji/Popularity/JSON/popDetails.json
(previous days are,e.g. popDetails-20100515.json) so others can make their own fancy graphs from that. In the future we may be able to access the popularity db directly as it won't make too much sense to replicate the database if we want to keep a history.
I added this to the wiki page for file access patterns. https://www.gridpp.ac.uk/wiki/FileAccessPatterns which covers both this kind of stuff and local file access issues.
Wahid
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
|