A Multi-Raft implementation in Rust.
- Supports Multi-Raft. Mutliple Raft processes can coexist in a single OS process so they can share resources efficiently.
- Tested with 1000 shards per node.
- Heartbeats in shards are batched to reduce the network overhead.
- Based on Tonic and efficient gRPC streaming is exploited in log replication and snapshot.
- Efficient backend implementation using redb.
- Writes in shards are batched in one transaction.
- Phi Accrual Failure Detector is used for leader failure detection.
- The adaptive algorithm allows you to not choose a fixed timeout number before deployment and to deploy Raft node in even geo-distributed environment where the latency between nodes isn't identical.
- phi-detector: Implementation of Phi Accrual Failure Detector in Rust.
Name: Akira Hayakawa
Email: ruby.wktk@gmail.com