Panasas File System

General remarks

The parallel Panasas file system went into operation in 2013 together with the btrzx3 cluster. The Panasas file system is attached to the clusters' High-Performance networks such that the nodes can write data in parallel in a very efficient way. One major advantage is that a number of nodes can even write or read a single, shared file in parallel, e.g., using MPI-IO, an therefore utilize the up to 20x 10 GBit/s I/O rate .

The total capacity of the file system is about 82 TByte, which is split into two blocks of 41 TByte each called data and scratch. While data is backuped and only available to specific groups, scratch is not backuped and free to use for anyone. Since the Panasas file system is intended to provide high IO rates, it is best practice to not use it for long-term storage.

Mount points

• btrzx3:
  /panasas/scratch
  /panasas/data
• btrzx2:
  /scratch
  /data
• btrzx4 (mounted as NFS):
  /scratch
  /data
• btrzx5 (mounted as NFS via btrzx3):
  /panasas/scratch
  /panasas/data