[HIVE-23522] repl bootstrap load: optimize partition loads

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: HiveServer2
• Assignee: Aasha Medhi
• Reporter: Anishek Agarwal
• Created: 21/May/20 06:02
• Updated: 21/May/20 06:25

Description

when "hive.repl.dump.metadata.only.for.external.table" = true is used in repl dump, we only dump metadata for external tables. partitioned external tables currently on the load side load one partition at a time, even though HMS has an api for bulk partition update. we should for such scenarios use the bulk api. this will significantly improve performance during bootstrap phase.