[HIVE-21354] Lock The Entire Table If Majority Of Partitions Are Locked

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 3.2.0, 4.0.0
• Component/s: HiveServer2
• Assignee: David Mollitor
• Reporter: David Mollitor
• Created: 28/Feb/19 16:18
• Updated: 26/Apr/21 14:50

Description

One of the bottlenecks of any Hive query is the ZooKeeper locking mechanism. When a Hive query interacts with a table which has a lot of partitions, this may put a lot of stress on the ZK system.

Please add a heuristic that works like this:

1. Count the number of partitions that a query is required to lock
2. Obtain the total number of partitions in the table
3. If the number of partitions accessed by the query is greater than or equal to half the total number of partitions, simply create one ZNode lock at the table level.

This would improve performance of many queries, but in particular, a select count(1) from table ... or ... select * from table limit 5 where the table has many partitions.

Issue Links

• depends upon: HIVE-21347 Store Partition Count in TBLS
• duplicates: HIVE-23714 Add new configuration for lock escalation
• relates to: HIVE-20890 ACID: Allow whole table ReadLocks to skip all partition locks