JProfiler Get rid of your performance problems and memory leaks! Modern Garbage Collection Tuning Java Performance Training Courses COURSES AVAILABLE NOW. We can provide training courses to handle all your Java performance needs Java Performance Tuning, 2nd ed The classic and most comprehensive book on tuning Java Java Performance Tuning Newsletter Your source of Java performance news. Subscribe now! Enter email: Training online Threading Essentials course JProfiler Get rid of your performance problems and memory leaks!

Back to newsletter 172 contents

For those of you interested in knowing what I do when I'm not producing these newsletters, I've described the Java team and technology at the fast growing startup I'm working at - I won't go into any more detail here, except to say I'm having a lot of fun.

A note from this newsletter's sponsor

In the linked to presentations, Emad Benjamin gives some lovely tips about sizing your microservices - from the "too small" angle, which is pretty timely, as all the current advice is about moving to smaller services from monolithic ones. He points out firstly, that for a given amount of heap, the more JVMs you split that heap over, the less overall efficiency you get - because two lots of GCs in two different JVMs is not as efficient as one lot of GCs in one JVM the same heap size as those two JVMs. Additionally, with multiple JVMs on the same set of cores, if GC execution happens to coincide across JVMs then you get contention on the CPU (GC is CPU intensive).

Finally, a more detailed point by Emad: if the majority of your requests need both microservice A and microservice B, and these two microservices both each need to scale to the same number of JVMs, there is no point at all in having those microservices running in separate JVMs - you may as well coalesce those microservices into the same JVM and gain the benefit of single JVM efficiency.

And one more presentation I've linked to this month, details from Netflix by Jeremy Edberg and Philip Fisher-Ogden, is an absolute must see presentation, covering so much ground on building resilient systems. I particularly liked the tip that you should design a microservice so it can run at least three copies across your system - if it works that way, you have overcome the major challenges in making that microservice scale horizontally.

Now on to all our usual sections: links to tools, articles, news, those talks and as ever, all the extracted tips from all of this month's referenced articles including those gems above.

A note from this newsletter's sponsor

News

Java performance tuning related news.

• JVM statistics cause garbage collection pauses
• Excelsior JET 10.5 includes a new better performing garbage collector called CoreBalance
• Java Flight Recorder can be enabled dynamically; and you can use -XX:+DebugNonSafepoints for any instrumentation that operates without safepoint pauses if you want more accurate metadata
• Profiling analysis technique which reduces the noise in profiles
• Prior art request for WO2014027990 "Performance tests in a continuous deployment pipeline"
• Benchmarking MpmcArrayQueue against ConcurentLinkedQueue

Tools

Java performance tuning related tools.

• Offheap Java POJOs with guaranteed memory alignment
• Facebook's now open sourced Augmented Traffic Control (ATC) tool to simulate network conditions
• technology for delivering real-time data
• OpenHFT Koloboke high performance collections

A note from this newsletter's sponsor

Articles

• Using wait(), notify() and notifyAll() in Java: common problems and mistakes (Neil Coffey/javamex)
• Why Non-Blocking? (Bozhidar Bozhanov/bozho)
• Understanding the JVM and Low Latency Applications (Simon Ritter/Oracle)
• Tuning Large Scale Java Platforms (Emad Benjamin, Jamie O'Meara/InfoQ)
• You Won't Believe How the Biggest Sites Build Scalable and Resilient Systems! (Jeremy Edberg, Philip Fisher-Ogden/InfoQ)
• ExecutorService vs ExecutorCompletionService in Java (Akhil Mittal/DZone)

Jack Shirazi

Back to newsletter 172 contents