| title | category |
|---|---|
TiKV Roadmap |
Roadmap |
This document defines the roadmap for TiKV development.
- Titan - Use the separated key-value engine in production
- Pluggable Engine Interface - Clean up the engine wrapper code and provide more extendibility
- Joint Consensus - Change multi members safely
- Quiescent Region - Reduce the heartbeat cost of inactive Regions
- Raft engine - Customize engine to replace RocksDB
- Remove KV RocksDB WAL - Use Raft log as the WAL of state machine engine
- Crossing datacenter optimization
- Raft witness role - Can only vote but not sync logs from Raft leader
- Follower snapshot - Get the snapshot from the follower when a new node is added
- Chain replication - Get Raft logs from the non-leader node
- Flow control - Do flow control in scheduler to avoid write stall in advance
- History version - Use another place to save history version data
- Performance
- Optimize storage format
- Reduce the latency of getting timestamp
- Optimize transaction conflicts
- More pushdown expressions support
- Common expression pipeline optimization
- Batch executor - Process multiple rows at a time
- Vectorized expression evaluation - Evaluate expressions by column
- Chunk - Compute over the data stored in the continuous memory
- Explain support - Evaluate the cost of execution
- Use Raft learner to support Backup/Restore/Replication
- Rust
- Go
- Java
- C++
- Visualization - Show the cluster status visually
- Learner scheduler - Schedule learner Regions to specified nodes
- Improve hot Region scheduler
- Improve simulator
- Range scheduler - Schedule Regions in the specified range
- Tolerate corrupted Region - TiKV can be started even when the data in some Regions is corrupted
- Support huge Region whose size is larger than 1 GB