Apache Ratis - A High Performance Raft Library
summary
Apache Ratis is a Java 8 library that implements Raft with concrete features: leader election, log replication, membership change and log compaction, pluggable RPC/state machine/log/metrics, TLS auth, snapshots and CLI commands for leader transfer and metadata edits. It targets data‑intensive systems (Apache Ozone, Apache IoTDB, Alluxio) and exposes async event‑driven APIs plus a streaming, zero‑copy pipeline that the implementation reports as about 3x performance higher than the simple append API; multi‑Raft groups and stale/read‑from‑follower modes are supported.
AI-written summary · about
chapters (16)
- 0:00Introduction and speaker background
- 1:20What is consensus and Raft basics
- 4:10Apache Ratis project origin and history
- 6:23General use cases for Raft
- 7:06Apache Ratis features overview
- 8:55Ratios high performance architecture
- 10:22Data replication and log management
- 11:28Streaming API and performance benefits
- 13:11Read consistency and stale reads
- 15:02Watch requests and replication levels
- 15:48Multi-Raft and server roles
- 16:08Security and TLS support
- 16:50Leadership management and snapshots
- 17:56Server priority and leader locality
- 18:39Listener members and failover
- 19:22Command interface and leader transfer
related talks
43mRa: a Raft implementation
55mImplementing Raft in RabbitMQ
01h 12mCo-designing Raft + thread-per-core execution model for the Kafka-API
47mDistributed Systems: When Democracy and Computer Science Come Together
02h 04mKafka with Raft in Production: A New Era for Distributed Commit Logs
41mThe Raft Consensus Algorithm
42mApache Ozone - State of the Union
44mCluster Consensus: When Aeron Met Raft