Bottlenecks are out! Java Cache Standard (JSR-107) is in!

Transcript

2. Bottlenecks are out!
   Java Cache Standard (JSR-107) is in! ! Galder Zamarreño Senior Software Engineer Red Hat, Inc ! 20th April 2013, São Paulo
3. Galder Zamarreño • R&D Engineer, Red Hat Inc. • Infinispan developer • Escalante founder and lead • Twitter: @galderz
4. Agenda • Why we care about bottlenecks • What is JCache? • Infinispan JCache implementation in action
5. Why do we care about bottlenecks?
6. Bottleneck problems Bottlenecks decrease throughput, increase latency... and make users :(((
7. We care because we want to make users :))
8. How can JCache help remove bottlenecks?
9. What is JCache? Temporary caching API for Java (driven by JSR-107), originally scheduled to be in EE7... but not any more :(
10. Infinispan JCache impl JCache 0.6 implemented in Infinispan 5.3.0.Alpha1, but some parts still in flux
11. Caching data Caching data that is expensive to retrieve (i.e. DB) or hard to calculate can boost performance :))
12. Demo 1: Caching data
13. Why not use a ConcurrentHashMap to cache data?
14. Beyond CHMs... JCache allows data to expire, is pluggable with persistent stores, is designed with distribution in mind, ...etc
15. Not a ConcurrentMap javax.cache.Cache does not extend j.u.c.ConcurrentMap, but borrows some ideas...
16. Demo 2: Cache vs ConcurrentMap
17. Scale up by distributing the cache
18. Go multi-node! When a single machine cannot cope with load, and not $$$ for more powerful machine...
19. Go multi-node! Run app in multiple nodes, and distribute cache contents!
20. Get failover for free! Providing failover for data stored in caches is one of the biggest advantages of distributed caches!
21. Demo 3: Cache goes distributed!
22. Integrate JCache with enterprise technologies
23. CDI and Transactions Integration with CDI or JTA Transactions is optional in the spec, but Infinispan implements it :)
24. Demo 4: Transactional JCache
25. Summary Q - How JCache can help you remove bottlenecks?
26. Summary • Provides in-memory, temporary storage, that's better than a (Concurrent) HashMap • Allows you to distribute the cache, to scale out your application! (and get failover!) • Makes it easy to integrate with enterprise technologies
27. Questions • http://infinispan.org • http://www.jcp.org/en/jsr/detail?id=107 • https://github.com/galderz/jcache