Continuous Performance Regression Testing with JfrUnit (P99Conf)

Functional unit and integration tests are a common practice to detect and prevent regressions within a software component or application's behavior.

Things look different, though, when it comes to performance-related aspects: how to identify an application is slower than it used to be? How to spot higher memory consumption than before? How to find out about sub-optimal SQL queries that sneaked in? Any performance tests based on metrics like wall-clock time or through-put are not portable. They are subject to a specific execution environment such as a developer laptop, CI, or production-like environment.

Welcome JfrUnit: based on the JDK Flight Recorder (JFR), it allows you to implement assertions based on all kinds of JFR events emitted by the JVM or your application. JfrUnit makes it very easy to identify potential performance issues by asserting metrics that may impact your application's performance, like an increased object allocation rate, retrieval of redundant data from the database, loading of unneeded classes, and much more.

Come and join us for this code-centric session to learn about:

* Using JDK Flight Recorder and JfrUnit for implementing performance regression tests
* Emitting JFR events from 3rd party libraries using JMC Agent
* Analyzing performance regressions in JDK Mission Control

Transcript

1. Brought to you by Continuous Performance Regression Testing with JfrUnit Gunnar Morling Principal Software Engineer at
2. Today’s Objective: Learn About... ▪ Implementing performance regression tests with JDK Flight Recorder and JfrUnit ▪ Analyzing performance regressions in JDK Mission Control ▪ Emitting JFR events from 3rd party libraries
3. Gunnar Morling Principal Software Engineer at Red Hat ▪ Debezium ▪ Quarkus ▪ JfrUnit, ModiTect, kcctl ▪ Spec Lead for Bean Validation 2.0 ▪ Java Champion ▪ @gunnarmorling
4. Challenges of Performance Tests ▪ Tests based on throughput or latency are dependent on specifics of the environment • Require production-like set-up • Impacted by unrelated concurrent load (e.g. on CI server)
5. JfrUnit — Assert Performance Impacting Metrics ▪ Assert proxy metrics like memory allocation or IO • Based on JDK Flight Recorder events • Failures may indicate a performance regression ▪ Plain unit tests ▪ Analyse in JDK Mission Control
6. Hello, JfrUnit!
7. Tools
8. OpenJDK Flight Recorder ▪ “Troubleshooting, monitoring, and profiling framework embedded into the JVM” ▪ 150+ built-in event types; custom ones ▪ Open-source since Java 11 (backported to 8) ▪ Low overhead, meant for “always-on” ▪ Accessible via recording files and streaming
9. JFR Event Types https://bestsolution-at.github.io/jfr-doc/
10. OpenJDK Mission Control ▪ Control, visualize, and analyze JFR recordings ▪ And more: • JMX console • heap dump analyzer • etc.
11. Demo Time!
12. JfrUnit — Discussion and Limitations ▪ Robust tests with very fast feedback cycle • Independent from environment • Proactive identification of issues ▪ Metrics need solid understanding; e.g. increased allocation may or may not be a problem ▪ Cannot identify all issues, e.g. locks showing up only under load ▪ Won’t help with issues elsewhere, e.g. a bad query execution plan
13. Wrap-Up
14. Take Aways ▪ JfrUnit: a tool in the performance testing box • Complement - no replacement - for other tests ▪ JMC Agent: produce events from code not under your control ▪ Move beyond Java 1.8 🚀
15. Outlook — Keeping Track of Historic Values https://horreum.hyperfoil.io/
16. Resources ▪ JfrUnit https://github.com/moditect/jfrunit ▪ Introduction to JfrUnit https://www.morling.dev/blog/towards-continuous-performance-regression-t esting/ ▪ Introduction to JMC Agent https://developers.redhat.com/blog/2020/10/29/collect-jdk-flight-recorder-ev ents-at-runtime-with-jmc-agent/ ▪ Profiling Java inside containers with ContainerJFR https://developers.redhat.com/devnation/tech-talks/containerjfr
17. Brought to you by Gunnar Morling gunnar@hibernate.org @gunnarmorling