Take a survey of scientific computing practitioners for a definition of HPC and you will undoubtedly end up with a broad picture encompassing various shapes, sizes and domains of individual and ensemble jobs. Ask about the storage needed to support these workloads and you’ll find they all need a high-performance filesystem (HPFS) designed for parallel access.
OpenStack is now a hot topic in the HPC world. Whether it’s using OpenStack as a bare-metal provisioning service or looking to offer a flexible, integrated IaaS close to primary HPC. However, the question of how to integrate a HPFS with OpenStack is not yet well answered. Little wonder it was one of the priority areas surfaced by the Scientific WG in Austin. In this talk architects and operators from Cambridge University in the UK and Monash University in Australia will explain their approaches to integrating Lustre and OpenStack, the tradeoffs involved, the gaps remaining, and share performance data.
- The different ways in which Lustre can be integrated within an OpenStack environment.
- Approaches for orchestrating integration of parallel filesystems into an OpenStack cloud.
- Security concerns and other issues that should be taken into consideration.
- How this use-case different from typical NAS (NFS, CIFS) and how it compares to new “software-defined” distributed storage systems such as Ceph
Registration Opening Soon
Registration will open for you to purchase passes for the OpenStack Summit very soon!
We're excited to see you there!
