Roadmap

Milestones Targeted for 1.0

• Hinted handoff
  • enables a cluster to be always writeable as long as one node is up.
• Incremental backups
  • Allow one cluster to incrementally be backed up to another cluster
    with similar partitioning constants
• Buffered writes
  • Enable a mode with reduced durability where writes are buffered and
    flushed to disk in the background
• Avg Latency and 99.9% latency within the same order of magnitude as
  those described in the paper
• Load distribution efficiency above 0.95 (mean load / max load)
• Integration test suite for distributed failure scenarios