SGER: The Chaotic Behavior of Automatic Memory Management

点击翻译按钮获取中文摘要

"Garbage collection" is the quaint name for automatic storage management, a feature common to many modern programming languages, notably Java. It has always been a cause of varying performance, though advances in garbage collection (GC) algorithms and tuning over the past two decades have made it tolerable for some production applications.Chaos, as popularized by the saying that the flapping of a butterfly's wings might determine whether a hurricane occurs halfway around the globe, is the study of complex non-linear systems. Chaotic systems are not random, just very complex, requiring different tools to analyze and understand them. Chaos also places bounds on what we can know or predict about a system, and on how we can control it.Over the last twenty years there was some evidence collected that GC, and computer systems in general, are chaotic. This SGER project will try to build evidence one way or the other. The implications are significant in terms of the predictability and controllability of this aspect of computer systems. However, the work is also significant as a focused step in understanding the possible chaotic nature of many aspects of computer systems, as a few researchers have begun to collect evidence.