| CPC G06F 3/0634 (2013.01) [G06F 3/0604 (2013.01); G06F 3/0653 (2013.01); G06F 3/067 (2013.01)] | 20 Claims |

|
1. A method of storage system reconfiguration alternation for bandwidth intensive Input/Output command (IO) workloads and processing intensive IO workloads, comprising:
monitoring an IO workload of a storage system for characteristics indicative of the onset of a primarily bandwidth intensive IO workload;
determining from the monitored IO workload whether the IO workload on the storage system is the primarily bandwidth intensive IO workload or is not the primarily bandwidth intensive IO workload;
in response to a determination that the IO workload is not the primarily bandwidth intensive IO workload, applying a first default set of configuration policies to the storage system to configure the storage system to respond to the primarily processing intensive IO workload; and
in response to a determination that the IO workload is the primarily bandwidth intensive IO workload, applying a second alternate set of configuration policies to the storage system to reconfigure the storage system to respond to the primarily bandwidth intensive IO workload;
wherein the first default set of configuration policies includes a default Least Recently Used (LRU) policy that specifies a default amount of time data should be retained in shared global memory after receipt of read IO operations requesting the data;
wherein the default amount of time is application specific, such that the default LRU policy separately specifies the default amount of time data should be retained in shared global memory for each individual application;
wherein the second alternate set of configuration policies includes an alternate LRU policy that specifies a second amount of time data should be retained in shared global memory after receipt of read IO operations requesting the data;
wherein the second amount of time is application specific, such that the default LRU policy separately specifies the second amount of time that data should be retained in shared global memory for each individual application; and
wherein the second amount of time is shorter than the default amount of time for at least some of the individual applications.
|