Chaos Minions: Harassing the Worlds Largest Supercomputer10/06/2017 Objective: Guidance As HPC systems increase in size and complexity, there is a growing need for resilience validation.Chaos Minions is a framework in which fault injections and recovery are combined into an automated solution.The framework runs harassers on targeted components and provides randomly generated harassment to the system. A workload can run parallel to the framework to identify bottlenecks in resiliency. Speaker(s)
|