• Support up to 200+ nodes in a cluster
    • Fast full backup and restoration
    • Dynamically split and merge hot spot Regions
    • Fine-grained memory control
    • Raft
      • Read-only Replicas
    • Improve scan performance
    • Dynamically increase the number of worker threads (WIP)
    • Flexibly increase read-only replicas (WIP)
    • Optimize the scheduling system to prevent QPS jitter
    • Refactor log content
    • Fast incremental backup and restoration
    • Flashback to any point-in-time
    • Hierarchical storage
    • Fine-grained QoS control
    • Raft
      • Chain replication of data
      • Witness role
    • Storage engine
      • Support splitting SSTables according to Guards During compaction in RocksDB (WIP)
      • Separate cold and hot data
    • Improve fast backup performance
    • Improve fast restoration performance
    • 1PC
    • Support storage class memory hardware