Techniques for identifying I/O workload patterns using key performance indicators

摘要

Techniques for identifying I/O workload patterns may include monitoring key performance indicators (KPIs) for a monitoring period; and at the end of the monitoring period, performing processing including: determining whether there have been a specified number of occurrences of a predefined event with respect to a data portion of the application, wherein the predefined event is a violation of an application KPI for the application and also a violation of at least one of a plurality of data storage KPIs for the data portion of the application; and responsive to determining that there have been the specified number of the occurrences of the predefined event with respect to the first data portion of the application, recording a pattern of I/O workload for the data portion in accordance with the occurrences of the predefined event. Hints, such as affecting data movement and/or compression, may be generated based on detected patterns.