[HBASE-1811] Snapshot HFile and region statistics at compaction time and make info available to clients

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Duplicate
• Assignee: Unassigned
• Reporter: Andrew Kyle Purtell
• Created: 02/Sep/09 23:17
• Updated: 11/Jun/22 21:02
• Resolved: 15/May/14 02:18

Description

Consider snapshotting HFile and region statistics at major and minor compaction time and making the info available to clients:

• Key statistics
  • cardinality
  • length avg/min/max/stdev
  • information content measure (entropy, etc.)
  • histogram
    etc.

• Value statistics
  • length avg/min/max/stdev
  • information content measure (entropy, etc.)
  • histogram
    etc.

• Region statistics
  • density estimation
  • KV count
  • total storage size (on disk)
  • total storage size (uncompressed)
    etc.

Issue Links

• is part of: HBASE-7958 Statistics per-column family per-region
• relates to: HBASE-1805 Cluster status and region topology visualization