Dataset splitting based on workload footprint analysis

摘要

Techniques are provided for splitting a computer dataset between multiple storage locations based on a workload footprint analysis of that dataset. As a computer accesses data storage, its input/output (I/O) access can be monitored, as well as a working dataset of that dataset. The I/O access patterns can be used to determine an application of the computer that is generating the I/O. The application and the working dataset can be used to determine a split for the dataset across multiple storage locations. The dataset can then be split according to the determined split.