Systems and methods for tracking working-set estimates with a limited resource budget

摘要

Embodiments of the systems and techniques described here can leverage several insights into the nature of workload access patterns and the working-set behavior to reduce the memory overheads. As a result, various embodiments make it feasible to maintain running estimates of a workload's cacheability in current storage systems with limited resources. For example, some embodiments provide for a method comprising estimating cacheability of a workload based on a first working-set size estimate generated from the workload over a first monitoring interval. Then, based on the cacheability of the workload, a workload cache size can be determined. A cache then can be dynamically allocated (e.g., change, possibly frequently, the cache allocation for the workload when the current allocation and the desired workload cache size differ), within a storage system for example, in accordance with the workload cache size.

主权项

1. A method for caching allocation performed on a storage system, the method comprising:
determining a working-set size for a first workload, wherein the working-set size is determined based on a number of unique blocks accessed over a monitoring interval comprising a number of Input/Output (IO) accesses; adjusting the working-set size by,
in response to detecting a change in the first workload, increasing the number of IO accesses in the monitoring interval; anddetermining the working-set size based on a number of unique blocks accessed over the increased number of IO accesses in the monitoring interval; estimating an amount of data suitable to be stored in cache for the first workload based on the adjusted working-set size; determining a workload cache size based on the amount of data suitable to be stored in cache for the first workload; allocating a cache within the storage system in accordance with the workload cache size; evaluating an impact of the workload cache size on a service level objective of a second workload with a higher priority; and reducing the workload cache size when a negative impact on the second workload is identified.