[IMPALA-5498] Support for partial sorts

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Impala 2.10.0
• Fix Version/s: Impala 2.10.0
• Component/s: Backend
• Labels: kudu performance
• Target Version: Impala 2.10.0
• Epic Color: ghx-label-6
• Assignee: Thomas Tauber-Marshall
• Reporter: Thomas Tauber-Marshall
• Created: 13/Jun/17 16:28
• Updated: 05/Dec/17 22:28
• Resolved: 24/Jul/17 14:31

Description

Impala's sorting code currently only allows for full sorts, but it could be extended to support partial sorts.

This would be useful in situations where the sorting is being done for performance rather than correctness. For example, a recent change (IMPALA-3742) sorts rows for an INSERT before sending them to Kudu. Doing only a partial sort could speed this up and reduce the memory required while still retaining the primary benefit of reducing the load on Kudu.

Issue Links

• breaks: IMPALA-6280 Invalid plan for sorted INSERT with an outer join and null checking