A Dynamic Adjustment Mechanism with Heuristic for Thread Pool in Middleware

Abstract

Thread pooling is an important technique of performance optimization in middleware. With the consideration of the features of Internet applications, the configuration of thread pool in middleware needs to be adjusted dynamically on the basis of perceiving the run-time context. However, how to find out effective influencing factors which make the adjustment to have better adaptability remains to be discussed further. The paper firstly presents a thread pool model in context of Web application server based on M/M/1/K/•/FCFS queuing system. A dynamic mechanism which imports certain of heuristic factors for reflecting the context at run-time is studied to adjust the size of thread pool so as to adapt to the changes of resources well. The prototypical experiments verify the effective influence of heuristic factors that exert on adjustment of thread pool size and show that the presented mechanism can be helpful for improving the performance of system.