Storage
The following points/suggestions should be considered when running jobs concurrently:
- HPC storage has been purchased to deliver expected capacity requirements - evaluation of vendor responses is dominated by $/TB.
- Our current (2021) general use storage platform is very much entry level capabilities when it comes to performance.
- Researchers should avoid running parallel/concurrent
tar,gzip,etc. commands. Utilities such as pbzip2 should also be avoided.- If your computational research is I/O intensive, ensure that it is configured to use local scratch space (
/fast) - Consider other researchers and create a separate, sequential operations job for all your post-job I/O transactions.
- If your computational research is I/O intensive, ensure that it is configured to use local scratch space (
- Compression tools don't provide much benefit when working on binary data files - it's generally better to leave binary files uncompressed
Performance versus block size (read/write)
The following output shows the performance achievable, on SATA SSDs, for reads/writes of differing block sizes.
Block size 4K Write: 256000000 bytes (256 MB, 244 MiB) copied, 2.97716 s, 86.0 MB/sBlock size 4K Read : 256000000 bytes (256 MB, 244 MiB) copied, 2.54667 s, 101 MB/sBlock size 64K Write: 255983616 bytes (256 MB, 244 MiB) copied, 0.287594 s, 890 MB/sBlock size 64K Read : 255983616 bytes (256 MB, 244 MiB) copied, 0.248797 s, 1.0 GB/s
The story told by the brief output above is universal - small files move around at much slower rates than large files.