High Performance Storage

/nobackup is a shared high performance storage system on the CAC systems that provides access to large amounts of disk for short periods of time at much higher speed than /home.

What is it good for?

/nobackup is best for large reads and writes of very large data files. Checkpoint/restart files and large datasets that are read from or written to frequently in the course of a run are good examples of wise uses of this space. Another good use of /nobackup is for code that uses MPI-IO provided by the MPI-2.0 standard (MPI-IO allows all processes to access a single file in parallel).

What is it bad for?

/nobackup is not as good for small reads and writes, compiling code, or long-term storage, although in some cases it may be faster than /home for small files. /nobackup does not count against your home directory quota but will be purged of old files if space becomes low. It is greatly to your advantage to carefully manage your space so we don't manage it for you.

Policies

Only data for currently running jobs should be left on /nobackup. The system is not backed up and is vulnerable to data loss.

Data should be cleaned by users often, old data will be removed to maintain working space. Important data should be moved to /home or AFS to avoid loss.

Details

The filesystem uses the Lustre cluster file system and scales to 100s of GB/s IO. The CAC's implementation is 2 metadata servers (for redundancy) and 2 storage servers (for redundancy, performance and capacity), each with 7 storage targets, providing a total usable storage space of 49TB. The storage servers are Sun x4500s with 48 1TB disks each (the extra disks are used for redundancy to improve reliability).