July 6, 2017 –
I/O performance for HPC: finding the right access pattern and avoiding interference
Scientific applications are executed in a high performance computing (HPC) environment, where a parallel file system (PFS) provides access to a shared storage infrastructure. The key characteristic of these systems is the use of multiple storage servers, from where data can be obtained by the clients in parallel. The performance observed by applications when accessing a PFS is directly affected by the way they perform this access, i.e. their access pattern.
In this seminar, I'll discuss my work with the Ondes3D seismic simulation, which was focused into changing the application's access pattern to improve I/O performance without changing the output format. Moreover, I'll discuss my previous and current work on I/O scheduling at different levels of the I/O stack, pointing current challenges for future work.