Hazelcast - Hitchhiker’s Guide. Grzegorz Piwowarek. Java Fest 2020 Online
summary
Hazelcast is more than a cache: it exposes entry processors to run logic where the data lives, a distributed executor service, and SQL-and-predicate queries; it offers a CP subsystem (Raft, needs 3+ members) for strong consistency, CRDTs like PN-counters for conflict-free merges, and HyperLogLog (approximate cardinality) for tiny-memory distinct counts. Cluster-wide Flake IDs guarantee unique IDs even across split-brain, Jet provides a separate streaming engine, HD/off-heap memory + hot-restart reduce GC pain, and enterprise adds hot-cache, WAN replication and managed cloud.
AI-written summary · about
chapters (26)
- 0:00Introduction
- 4:59What is Hazelcast
- 13:17How to use it
- 15:07Ecosystem
- 17:20Entry Processor
- 19:16Executor Service
- 22:47CP Theorem
- 25:30CP Subsystem
- 28:03Advanced Data Structures
- 30:28Unique IDs
- 32:17Conflict Free Replication
- 33:25Integrations
- 37:53Iwata
- 39:55Hazelcast as Cloud
- 40:35Hot Cash
- 41:33Automatic failover
- 42:11Multiple replication
- 42:52HD Memory
- 44:26DC Memory
- 44:55SQL Engine
- 45:43Circle Jet
- 47:28Pipelines
- 48:06Architecture Design Documents
- 49:13Training
- 49:59Help
- 50:48Conclusion
related talks
05mStartup Showcase: Hazelcast
59mDistributed applications using Hazelcast
30mDistributed Computing with Hazelcast
45mEasy Scaling with Open Source Data Structures
34mHigh Performance Cloud-Native Microservices
41mWhat can go wrong with the distributed system - the Hazelcast exp...
38mHazelcast That Could: Getting Lightning-Fast Responses at Hundreds of RPS
45mjava.util.concurrent for Distributed Coordination