Lessons from capacity planning a Java enterprise application: how to keep capacity predictions on target and cut CPU usage by 5x

摘要

Java applications are ubiquitous in mission-critical enterprise and online settings. Java automatic memory management brings safety and development efficiency benefits, however at the same time adds complexity for the capacity manager in terms of how to effectively predict the capacity and manage the efficiency of such environments. We describe actionable methodologies and key metrics that enabled us to highlight the hidden bottleneck of many Java applications; devise a business-oriented capacity model that represents Java memory bottlenecks: detect poor memory usage patterns and anticipate memory leaks; determine that the garbage collector drives the CPU usage of your servers; and show how the garbage collector might be your first scalability bottleneck.