SmartGC: Online memory management prediction for PaaS cloud models

Abstract

In Platform-as-a-Service clouds (public and private) an efficient resource management of several managed runtimes involves limiting the heap size of some VMs so that extra memory can be assigned to higher priority workloads. However, this should not be done in an application-oblivious way because performance degradation must be minimized. Also, each tenant tends to repeat the execution of applications with similar memory-usage patterns, giving opportunity to reuse parameters known to work well for a given workload. This paper presents SmartGC, a system to determine, at runtime, the best values for critical heap management parameters of JVMs. SmartGC comprises two main phases: (1) a training phase where it collects, with different heap resizing policies, representative execution metrics during the lifespan of a workload; and (2) an execution phase where it matches the execution parameters of new workloads against those of already seen workloads, and enforces the best heap resizing policy. Distinctly from other works, this is done without a previous analysis of unknown workloads. Using representative applications, we show that our approach can lead to memory savings, even when compared with a state-of-the-art virtual machine - OpenJDK. Furthermore, we show that we can predict with high accuracy the best heap policy in a relatively short period of time and with a negligible runtime overhead. Although we focus on the heap resizing, this same approach could also be used to adapt other parameters or even the GC algorithm.