[CASSANDRA-6142] Remove multithreaded compaction (and precompactedrow)

• Type: Bug
• Status: Resolved
• Priority: Low
• Resolution: Fixed
• Fix Version/s: 2.1 beta1
• Severity: Low
• Assignee: Jonathan Ellis
• Reporter: Jonathan Ellis

Authors

Jonathan Ellis

Reviewers

Marcus Eriksson

• Created: 03/Oct/13 15:38
• Updated: 16/Apr/19 09:32
• Resolved: 30/Oct/13 15:20

Description

There is at best a very small sweet spot for multithreaded compaction (ParallelCompactionIterable). For large rows, we stall the pipeline and fall back to a single LCR pass. For small rows, the overhead of the coordination outweighs the benefits of parallelization (45s to compact 2x1M stress rows with multithreading enabled, vs 35 with it disabled).

Issue Links

• is duplicated by: CASSANDRA-7249 Too many threads associated with parallel compaction
• is duplicated by: CASSANDRA-6118 Kill PrecompactedRow now that LCR is single-pass?
• is duplicated by: CASSANDRA-4182 multithreaded compaction very slow with large single data file and a few tiny data files
• is related to: CASSANDRA-7428 clean out DD.inMemoryCompactionLimit