Adaptive Calling of Garbage Collector to Decrease Memory Leaks Based on Methods Ranking

Background: Memory management with dynamic allocation one of the most complex problems for software managed environment.Memory leak problem usually appear with long running information server applications.Materials and Methods: Garbage collectors is an effective tool to manage memory leak but only when the amount of lost storage is bounded.It has been shown that application performance with garbage collector is highly dependent on the application behavior and resource availability.Meanwhile, the PageRank algorithm is a widely used scoring function of networks in general and of the World Wide Web graph in particular.Current approaches rarely considered the application specific methodology to handle the memory management problem.Results: In this study, we focused on decreasing the memory leak for large-scale information systems running on servers.An approach proposed to apply adaptive calling of garbage collector within the methods relevant to memory leak during the application running, where PageRank algorithm is adjusted to estimate the importance of each method.Conclusion: Java benchmark tool "DaCapo" has been used to test and compare the memory and processing time before and after applying the adjusted algorithm, which showed the memory leak ration is successfully decreased with minor effect on processing time.