apache / apache/datafusion

Defer task spawning in SortPreservingMergeExec to first poll

Open
#21,329 0 comments 1 reaction 0 assignees View on GitHub
enhancement
Dominant language
Rust
Stars
9.3k
Forks
2.4k
Avg merge
3d 7h
Merged PRs (30d)
344

Description

### Is your feature request related to a problem or challenge?

SortPreservingMergeExec::execute() eagerly calls execute() on all input partitions and spawns buffered tasks immediately, before the output stream is ever polled. This means resources are allocated and work
begins even if the stream is never consumed (e.g. query cancelled before first poll), and creates an unnecessary burst of concurrent tasks when many SortPreservingMergeExec nodes exist in a plan.

### Describe the solution you'd like

Defer the spawning of input partition tasks and construction of the streaming merge to the first poll_next() call on the output stream, rather than doing it eagerly in execute(). This can be done with a
wrapper stream that holds the initialization state and transitions from Pending to Running on first poll. The single-partition and zero-partition fast paths can remain unchanged.

### Describe alternatives you've considered

_No response_

### Additional context

_No response_

Contributor guide

Open the contributing guide

Research direction

Start at SortPreservingMergeExec::execute and trace how input partition tasks and the streaming merge are created before the output stream is polled. Verify the zero- and single-partition fast paths, then confirm that multi-partition work and merge construction begin on the first poll_next call and not before.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
data-engineering
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
65/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.