[CASSANDRA-11206] Support large partitions on the 3.0 sstable format

• Type: Improvement
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 3.6
• Component/s: Legacy/Local Write-Read Paths
• Impacts: Docs
• Assignee: Robert Stupp
• Reporter: Jonathan Ellis

Authors

Robert Stupp

Reviewers

T Jake Luciani

• Created: 22/Feb/16 17:14
• Updated: 1 week ago 20:22
• Resolved: 21/Apr/16 14:51

Description

Cassandra saves a sample of IndexInfo objects that store the offset within each partition of every 64KB (by default) range of rows. To find a row, we binary search this sample, then scan the partition of the appropriate range.

The problem is that this scales poorly as partitions grow: on a cache miss, we deserialize the entire set of IndexInfo, which both creates a lot of GC overhead (as noted in CASSANDRA-9754) but is also non-negligible i/o activity (relative to reading a single 64KB row range) as partitions get truly large.

We introduced an "offset map" in CASSANDRA-10314 that allows us to perform the IndexInfo bsearch while only deserializing IndexInfo that we need to compare against, i.e. log(N) deserializations.

Issue Links

• blocks: CASSANDRA-11369 Re-use IndexedEntry / IndexInfo after 11206
• is depended upon by: CASSANDRA-9738 Migrate key-cache to be fully off-heap
• is depended upon by: CASSANDRA-11584 Add stats about index-entries to per sstable-stats
• is related to: CASSANDRA-19557 ShallowIndexedEntry scenario: the same IndexInfo is read multiple times, per every read row
• relates to: CASSANDRA-11765 dtest failure in upgrade_tests.upgrade_through_versions_test.ProtoV3Upgrade_AllVersions_Skips_3_0_x_EndsAt_Trunk_HEAD.rolling_upgrade_test
• relates to: CASSANDRA-9754 Make index info heap friendly for large CQL partitions
• supercedes: CASSANDRA-10320 Build off-heap key-cache compatible API
• supercedes: CASSANDRA-16904 Check if size of object being added to RowCache and KeyCache is bigger than cache capacity