Fault Tolerance of HPC systems
Abstract
Cloud computing offers new capacity and flexibility solution to high performance computing applications with provisioning of great number of virtual machines for computational intensive applications. Fault tolerance allows HPC system on cloud with multiple of nodes to complete execution of computational intensive applications (HPC applications) in the present of fault. The most commonly used fault tolerance techniques for HPC is checkpoint/restart. However, checkpoint/restart increases the wall clock time of the execution of the computational intensive applications which increases the cost of running the application. We want to develop middleware that will reduce the execution time of computational intensive applications in the presence of fault
Intellectual Merit
The activities will create a low latency middleware that will reduce the impact of failure on computational intensive applications running on HPC systems
Broader Impact
Applications running on HPC systems will greatly benefit because most the computional intensive applications are scientific application which are usually affected by failure/fault
Use of FutureGrid
For my research - Currently, I am a PhD student
Scale Of Use
I will need to provision across different systems/nodes for VMs for an experiment