London Java Community - An Experiment in Continuous Deployment of JVM applications
Download as PPTX, PDF0 likes187 views
The document discusses continuous deployment of JVM applications utilizing tools like Hazelcast IMDG and Jet for in-memory processing and low-latency operations. It outlines traditional deployment challenges and introduces modern approaches like blue-green deployment and Kubernetes rolling updates. The presentation details Java agents, instrumentation APIs, and dynamic loading methods for modifying bytecode without downtime, while also highlighting areas for improvement in the deployment process.
London Java Community - An Experiment in Continuous Deployment of JVM applications
- 1. @nicolas_frankel An Experiment in Continuous Deployment of JVM applications Nicolas Fränkel
- 2. @nicolas_frankel Me, myself and I Former developer, team lead, architect, blah-blah Developer Advocate Lots of crazy creative ideas
- 3. @nicolas_frankel Hazelcast HAZELCAST IMDG is an operational, in-memory, distributed computing platform that manages data using in-memory storage and performs execution for breakthrough and scale. HAZELCAST JET is the ultra fast, application embeddable, 3rd generation stream processing engine for low latency batch and stream processing.
- 4. @nicolas_frankel Continuous Integration Compilation Unit testing Packaging Integration testing Containerization Storing in a registry Etc.
- 5. @nicolas_frankel Continous Stuff Continuous Integration • Should be the norm Continuous Delivery • Much harder Continous Deployment • No downtime
- 6. @nicolas_frankel Traditional JVM deployment model Stop the JVM Overwrite JAR / WAR Start the JVM again
- 8. @nicolas_frankel Continuous Deployment approaches Kubernetes Rolling Updated 1.0 1.01.02.0 2.0 2.0
- 9. @nicolas_frankel Change bytecode inside a JVM with no downtime?
- 10. @nicolas_frankel Plan 1. The Instrumentation API 2. The Attach API 3. Binding everything together with streaming 4. Demo!
- 11. @nicolas_frankel Java agents JAR with specific meta-data Part of Instrumentation API • Change bytecode of running applications
- 12. @nicolas_frankel Types of agent Static • Set on at startup • java -jar -javaagent:/path/to/agent.jar my.jar Dynamic • Loaded by the Attach API
- 13. @nicolas_frankel Agent configuration Static Dynamic Entry- point public static void premain( String agentArgs, Instrumentation inst) public static void agentmain( String agentArgs, Instrumentation inst) Manifest entry Premain-Class Agent-Class
- 14. @nicolas_frankel Instrumentation use-cases Change existing class implementation • With some limitations Apply automatic transformation of bytecode
- 16. @nicolas_frankel The Attach API Allows to dynamically load an agent inside another running JVM • No need to run in debug Requirements: 1. Need to run on the same system 2. Must know the PID of the JVM
- 18. @nicolas_frankel Options to move bytecode around Batching Streaming
- 19. @nicolas_frankel In-memory stream processing engines Cloud • Amazon Kinesis • IBM Streams On-premise • Apache Flink • Hazelcast Jet • Apache Beam ⁃ Abstraction over some of the above
- 20. @nicolas_frankel Hazelcast Jet Apache 2 Open Source Single JAR Unified batch/streaming API
- 21. Hazelcast In-Memory Computing Platform Ingest Combine Stream Compute Transform Publish In-Memory Operational Storage Join, Enrich, Group, Aggregate Windowing, Event Time Processing Distributed and Parallel Computations Filter, Clean, Convert In-Memory Subscriber Notifications Databases Kafka IoT Custom Connector Enterprise Applications Hazelcast IMDG File Watcher Socket Database Events Kafka Alerts Enterprise Applications Interactive Analytics Hazelcast IMDG, Maps, Caches, Lists Databases HDFS File
- 24. @nicolas_frankel
- 25. @nicolas_frankel Limitations of the demo No version tagging No loading of new classes redefineClasses() “must not add, remove or rename fields or methods, change the signatures of methods, or change inheritance”
- 26. @nicolas_frankel Improvements 1. Stream from the CI pipeline 2. Add additional meta-data in each class • e.g. from Git tag 3. Implement one’s custom classloader à la Spring DevTools
- 27. @nicolas_frankel Thanks for your attention! https://blog.frankel.ch/ @nicolas_frankel https://bit.ly/bytecode-streaming