An Automatic Adjustment Approach of Thread Quantity to Optimize Resource Usage

ABSTRACT

In order to achieve optimization of resource usage for servers, a method is designed to automatically adjustment approach of thread quantity when multiple types of tasks are running, which can effectively and automatically optimize resource utilization without manual intervention. It first monitors different kinds of resources and trains resource usage of servers when adding a thread for a type of task. It then dynamically adds or reduces thread quantity to adapt to the scenario of dynamic change in resource usage according to monitoring results. When resources are idle and thread quantity needs to be added, the problem of determining thread quantity is abstracted by multi-dimensional container loading problem and we propose a heuristic algorithm based on similar ratio which can quickly obtain thread quantity. The proposed algorithm not only avoids the work of setting thread quantity for parallel tasks, but also improves the utilization rate of CPU, I/O and throughput.

Keywords: