Kirk Pepperdine works as an independent consultant offering Java performance-training and consulting services. Prior to focusing on Java, Kirk developed and tuned systems written in C/C++, Smalltalk, and a variety of other languages. Kirk has written many articles and spoken at several conferences on the subject of performance tuning. He helped evolve http://www.javaperformancetuning.com as a resource for performance tuning tips and information.
The Garbage First Garbage Collector is touted as beng the collector of the future. Though it's suppose to more self tuning than any previous collector it still requires that you provide it with a number of parameters. Each of these parameters is used to affect different aspects of collector function and internal memory pool sizing. This session will cover how the internal memory pools are structured and how the collector functions. The session will also cover G1 GC logs and how to use that information to understand how to configure the G1.
If you wanna know what your garbage collector is up to the best place to look is in the GC logs. This session will explore what information can be can read from the logs and what you can do to improve GC performance. GC logs are like any other ascii log file in that they can be full of 1000s of cryptic entries which make trying to figure out what the logs are saying without any tooling, a daunting task. In this session I will introduce tooling designed to help you make sense out of your log files. Some of what the tooling can tell you is if your JVM's heap is misconfigured as is the case for a large number of applications currently running in products systems. These application are starved for memory in at least one of their internal memory pools. But it’s not only starvation that can a cause of poor performance or long GC pauses. Too much memory can be equally as problematic. Or your problem could be as simple as your application calling System.gc(). Properly read, GC logs are key to solving all your Java heap rooted performance problems.