US 12,124,969 B2
Determining whether to process a host request by training a machine learning module
Matthew R. Craig, Sahuarita, AZ (US); Beth A. Peterson, Tucson, AZ (US); Lokesh M. Gupta, Tucson, AZ (US); and Kevin J. Ash, Tucson, AZ (US)
Assigned to INTERNATIONAL BUSINESS MACHINES CORPORATION, Armonk, NY (US)
Filed by INTERNATIONAL BUSINESS MACHINES CORPORATION, Armonk, NY (US)
Filed on Jan. 23, 2019, as Appl. No. 16/255,816.
Prior Publication US 2020/0234159 A1, Jul. 23, 2020
Int. Cl. G06N 5/04 (2023.01); G06N 20/00 (2019.01)
CPC G06N 5/04 (2013.01) [G06N 20/00 (2019.01)] 24 Claims
OG exemplary drawing
 
1. A computer program product for determining whether to process in a system a host request from a host system, comprising a computer readable storage medium having computer readable program code embodied therein that when executed performs operations, the operations comprising:
processing, at a processing node of the system managing access to a storage, a host request from the host system, wherein the host request comprises a member of a set of host requests consisting of an input/output request to the storage, a query, a reserve command, and a recovery operation;
providing a machine learning algorithm that receives, as input information, information on the host request and information on processes executing in the system, and produces, based on the input information, output for the host request indicating a degree to which processing of the host request will have a negative impact on system performance;
determining output produced from the machine learning algorithm based on the input information for a received host request;
assigning a task to process the received host request in response to the output being less than a threshold negative impact;
determining an expected output as a function of a measured system resource usage and an optimum system resource usage;
determining a margin of error from the expected output and the output produced from the machine learning algorithm;
retraining the machine learning algorithm with the input information and the margin of error; and
providing the retrained machine learning algorithm to process a received host request at the host system in response to determining the output of the retrained machine learning algorithm indicates to process the received host request and to indicate not processing the host request in response to determining the output of the machine learning algorithm indicates to not process the received host request.