Intelligently responding to hardware failures so as to optimize system performance

摘要

A method, system and computer program product for intelligently responding to hardware failures so as to optimize system performance. An administrative server monitors the utilization of the hardware as well as the software components running on the hardware to assess a context of the software components running on the hardware. Upon detecting a hardware failure, the administrative server analyzes the hardware failure to determine the type of hardware failure and analyzes the properties of the workload running on the failed hardware. The administrative server then responds to the detected hardware failure based on various factors, including the type of the hardware failure, the properties of the workload running on the failed hardware and the context of the software running on the failed hardware. In this manner, by taking into consideration such factors in responding to the detected hardware failure, a more intelligent response is provided that optimizes system performance.

主权项

1. A method for intelligently responding to hardware failures so as to optimize system performance, the method comprising:
monitoring utilization of hardware in a system; monitoring software components running on said hardware in said system to assess a context of said software components running on said hardware; detecting a hardware failure; analyzing said hardware failure to determine a type of said hardware failure; analyzing properties of a workload running on said failed hardware, wherein said workload running on said failed hardware comprises an amount of processing that said failed hardware has been given to do at a given time, wherein said properties of said workload running on said hardware comprise creating an additional image of a pattern of said workload running on said hardware and placing said image of said pattern of said workload in a hot-standby mode in response to said hardware appearing likely to fan; and responding, by a processor, to said detected hardware failure based on said type of said hardware failure, said properties of said workload running on said failed hardware, and said context of said software running on said failed hardware.