JCache - It's finally here
2 likes1,139 views
This document provides an overview of JCache (JSR107), the Java standard for caching. It discusses why caching is used to improve performance, introduces the key concepts and classes in JCache, and covers how to use caches, entry processors, listeners, loaders/writers, and annotations with JCache. It also discusses future directions for JCache like asynchronous APIs and integration with Java 8 features.
JCache - It's finally here
- 1. JCache (JSR107) David Brimley The standard for Java Caching
- 2. Who is this guy? Senior Solutions Architect @ Hazelcast Java Developer for 17 years @dbrimley In Memory Data Grids for 8 years Worked for Investment Banks Started out coding Basic on a Commodore Vic-20 My fav computer was my old BBC Model B Long suffering Spurs fan. HSBC, Nomura, Credit Suisse, BZW, RBoS Oracle Coherence 4 Years Pivotal Gemfire 4 Years David Brimley Runs HUGL (Hazelcast User Group London) Low latency trading systems
- 3. What we’ll cover… • Why Caching? • Introduction to JCache. • Using Caches. • Entry Processors. • Listeners. • Annotations. • Missing Bits. • Questions.
- 4. Who uses caching? Quick Show of hands
- 5. Why Caching?
- 6. Why Caching? The Speed Factor (Memory .v. Disk) L1 cache reference 0.5 ns Branch mispredict 5.0 ns L2 cache reference 7.0 ns Mutex lock/unlock 25.00 ns Main memory reference 100.00 ns Compress 1k bytes with ZupZ 3,000.00 ns Send 1k bytes over 1Gpbs network 10,000.00 ns 0.01 ms Read 4k randomly from SSD 150,000.00 ns 0.15 ms Read 1 MB sequentially from memory 250,000.00 ns 0.25 ms Round trip within same datacenter 500,000.00 ns 0.50 ms Read 1 MB sequentially from SSD 1,000,000.00 ns 1.00 ms Disk seek 10,000,000.00 ns 10.00 ms Read 1 MB sequentially from disk 20,000,000.00 ns 20.00 ms Send packet CA->Netherlands->CA 150,000,000.00 ns 150.00 ms https://gist.github.com/jboner/2841832
- 7. Why Caching? Graph of Memory Prices Decreasing with Time (1957-2014) http://www.jcmit.com/mem2014.htm The Cost Factor
- 8. Why Caching? Lots of Use Cases Slow Database HTTP Proxy Browser Persistence Framework Disk Application Hard to scale Database Lots of Cache Types Local Cache Data Grid Document Store JPA Second Level JPA First Level
- 10. Introduction to JCache • Caching for the Java Platform • Produced via the JCP / JSR107 • Ratified March 2014 • 10 years in the making. • Works with SDK. • Spec Leads: Greg Luck, Brian Oliver
- 11. java.util.Map (Java 6/7) Key-Value Based API Supports Atomic Updates Entries Don’t Expire Entries Aren’t Evicted Entries Stored On-Heap Store-By-Reference https://gist.github.com/jboner/2841832 Introduction to JCache Map .v. Cache javax.cache.Cache (Java 6) Key-Value Based API Supports Atomic Updates Entries May Expire Entries May Be Evicted Entries Stored Anyware (ie: topologies) Store-By-Value and Store-By-Reference Supports Integration (ie: Loaders/Writers) Supports Observation (ie: Listeners Entry Processors Statistics
- 12. JCache makes NO assumptions about topology. Single Process or Distributed Introduction to JCache
- 13. Introduction to JCache • JCP Project http://jcp.org/en/jsr/detail?id=107 • Source Code https://github.com/jsr107 • Forum https://groups.google.com/forum/? fromgroups#!forum/jsr107
- 14. JSR107 TCK Compliant Vendors Introduction to JCache https://jcp.org/aboutJava/communityprocess/implementations/jsr107/index.html I pity the fool that claims to support JCache without TCK compliance As of 5th March 2015
- 15. Introduction to JCache Something to remember when comparing vendors of distributed caches… Do NOT Benchmark with a single instance!
- 16. Using Caches
- 18. Using Caches Classes Caching “service loader” CachingProvider “SPI Implementation” CacheManager “manager of caches” Cache “interface to a Cache”
- 20. Entry Processors
- 21. Entry Processors Compute in place • Atomic Operations in situ of data. • Eliminate Round-Trips (if distributed) • Lock Free • Needs Serialisation (if distributed) Application Application Cache Cache
- 23. Listeners
- 24. Listeners Callbacks for events in Cache • Follows Observer Pattern • Interface per event type • Created, Updated, Removed, Expired • Iterations of Events passed back • MutableCacheEntryListenerConfiguration
- 27. Cache Loaders / Writer Cache Misses / Cache Persist Application Cache Backing Store
- 28. Cache Loaders / Writer Classes Code Walkthrough
- 29. Annotations
- 30. Annotations Spring 4.1, Guice, CDI • @CacheResult • @CachePut • @CacheRemove • @CacheRemoveAll
- 32. Missing Bits
- 33. Missing Bits The Future of JCache • Async API • Query • JEE 8 • Servlet 4.0 (Session Cache) • Java 8 Lang Features (Lambda/Streams)
- 34. Questions ?
- 36. HAZELCAST • Free London Training • Hazelcast User Group • Visit our Booth Visit our Booth 5th Floor Get Free Training (Canary Wharf, Thursday 12th March) http://hazelcast.com/events/ Join the Hazelcast User Group London (HUGL) http://www.meetup.com/Hazelcast-User-Group-London-HUGL/ Download Hazelcast, the leading open source IMDG http://www.hazelcast.org