AFS for (Andrew File System) is a distributed file system accessible over the whole Computing Centre computers. AFS is based on a local cache technique on each computers where your data in read and write mode goes through. AFS manages the synchronization between your local data and the AFS files servers.
We recommend the use AFS where you have a reasonable number of files to manage and almost of small size. This come from 3 points:
The local cache mechanism with a size smaller than 1 GB on each computers. If you have a larger file, you spend your time to flush the cache with the server data.
The partition system management with a limited size. You will have a limitation on the volume of data you want to store on a directory.
The small network performance that gives a limitation in case of heavy flow and intensive acces to a large volume of data.
Nevertheless, AFS is useful in case of code, script edition, ... where you do not need high network performances and you do not manage a huge volume of data because the access to your data is done like on a local file system. Some AFS space are saved on our bakup system.
In case of problems, please submit a ticket at CC-IN2P3 User Support.