“I created a filesystem, but it is not performing as expected. How can the performance be improved?”
Originally posted at the Oracle University Learning Community
Unlike Block Volumes, File Systems adjust their performance dynamically (no tiers), and this can give the impression that the performance is low.
Most of the time, the issue is that the way it works is not fully understood. The aspects that need to be observed:
Available bandwidth - make sure that the instance has enough bandwidth allocated. VMs have bandwidth based on the number of OCPUs
Latency - file systems are contained in an availability domain but can be accessed from anywhere (if rules allow). Placing the instance in the same AD will minimize latency and improve performance.
Mount options - not setting window size for read and write operations, the client and server can negotiate values that provide the best performance.
Filesystem size - the available read and write throughput increases proportionally to the size of the file system.
Parallelism - multiple threads, multiple clients, and multiple mount targets. In particular, scalability will be greatest when clients and threads access independent portions of the file system. Also, use fss-parallel-tools.
Observing these items, it is easy to improve and maximize the performance. Do you have another tip? Let us know and share in the comments.