Introducing Infinispan
1 like2,634 views
This document introduces Infinispan, an open source distributed in-memory data grid. It provides high availability and elasticity through data replication across nodes. The API allows developers to treat Infinispan as a distributed concurrent hash map. Key features include transactions, persistence, querying, and map/reduce capabilities. Infinispan can be used as a local cache, cluster of caches, or autonomous data store accessed via protocols like REST and Memcached. Future plans include improvements to transactions, non-blocking state transfers, and support for cross-datacenter replication and eventual consistency.
Introducing Infinispan
- 1. Introducing Infinispan • Mircea Markus • Senior Software Engineer • Red Hat, Inc
- 2. Mircea Markus • R&D Engineer, Red Hat Inc. • Twitter: @mirceamarkus • Blog: mirceamarkus.blogspot.com
- 3. Agenda • What is Infinispan • API • Key features • Three use cases • The path ahead
- 4. What is Infinispan? • Distributed, in memory, data structure • Highly available • Elastic • Open source
- 6. High availability • Memory is volatile • Make redundant copies • Total replication (Replication Mode) • Partial replication (Distribution Mode) • Topology changes • Node will crash! • Re-arrange state
- 7. Elasticity • Expect • Node additions • Node removals • Topology changes • are totally consistent • do not "stop the world"
- 8. API • Key-value store • java.util.concurrent.ConcurrentHashMap • JSR-107 compliant • CDI support
- 9. Key features • Transactions • Persistence • Map/Reduce • Querying
- 10. Transactions • JTA transactions support • XA or Synchronization based enlistment • recovery! • Local transactions (batching) • Transactional modes • optimistic • pessimistic
- 11. Persistence/Cache Store • Used for • durability • increased storage • warm caches • Various implementations • jdbc, bdbje, file, cloud, remote/infinispan, cassandra • Extensible
- 13. Map/Reduce
- 14. Map/Reduce
- 15. Querying • What’s in C7? • Where is the white king?
- 16. Querying • Based on Hibernate Search/Lucene • Objects in the grid are indexed • index management
- 17. Use cases • Local cache • Cluster of caches • Autonomous data store • access protocols
- 18. Local cache
- 19. Features of a local cache • Eviction • Expiry • Write through, write behind • Preloading • Notifications • Statistics
- 20. Local cache not good enough...
- 22. Some limitations still • Client is affected by cache topology changes • Shared resources • Tier management • incompatible JVM tuning • security • garbage collection • Non-JVM clients
- 23. Cache servers
- 24. Client/Server protocols • REST • Memcached • Hotrod • proprietary • java, python, ruby
- 25. The path ahead • 5.2 • transactions performance • non-blocking state transfer • hotrod enhancements • 6+ • X-datacenter replication • eventual consistency • continuous queries
- 26. Thank you! • infinispan.blogspot.com • @infinispan • mirceamarkus.blogspot.com • @mirceamarkus