Horton

Re: [PATCH] Incremental sort (was: PoC: Partial sort)

Tomas Vondra <tomas.vondra@2ndquadrant.com>

View thread View on postgr.es
From: Tomas Vondra <tomas.vondra@2ndquadrant.com>
To: James Coleman <jtc331@gmail.com>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Alvaro Herrera <alvherre@2ndquadrant.com>, Michael Paquier <michael@paquier.xyz>, Rafia Sabih <rafia.pghackers@gmail.com>, Peter Geoghegan <pg@bowt.ie>, Simon Riggs <simon@2ndquadrant.com>, Shaun Thomas <shaun.thomas@2ndquadrant.com>, Dmitry Dolgov <9erthalion6@gmail.com>, Alexander Korotkov <a.korotkov@postgrespro.ru>, PostgreSQL Developers <pgsql-hackers@lists.postgresql.org>, Andreas Karlsson <andreas@proxel.se>
Date: 2020-04-06T19:57:22Z
Lists: pgsql-hackers

Attachments

Hi,

I've pushed the fist part of this patch series - I've reorganized it a
bit by moving the add_partial_path changes to the end. That way I've
been able to add regression test demonstrating impact of the change on
plans involving incremental sort nodes (which wouldn't be possible when
committing the add_partial_path first). I'll wait a bit before pushing
the two additional parts, so that if something fails we know which bit
caused it.

I've been running extensive benchmarks with the aim to detect any
regressions caused by this patch, particularly during planning. Attached
is the script I've used and spreadsheet with results. The numbers show
throughput with different queries (SELECT, EXPLANN and joins) with the
patches committed one by one. There's quite a bit of noise, even though
the script pins processes to cores and restricts CPU frequency. Overall,
I don't see any obvious regression - the numbers are generally within
0.5% of the master, and the behavior is the same even with -M prepared
which should not be subject to any planner overhead. I do have results
from another machine (2-socket Xeon) but the results are much more
noisy, although the general conclusions are about the same.


regards

-- 
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Commits

  1. Further adjustments to Hashagg EXPLAIN ANALYZE output

  2. Rework EXPLAIN format for incremental sort

  3. Fix typos and improve incremental sort comments

  4. Stabilize incremental_sort tests

  5. Minor improvements in Incremental Sort explain

  6. Consider Incremental Sort paths at additional places

  7. Fix representation of SORT_TYPE_STILL_IN_PROGRESS.

  8. Fix failures in incremental_sort due to number of workers

  9. Fix show_incremental_sort_info with force_parallel_mode

  10. Implement Incremental Sort

  11. Fix handling of "Subplans Removed" field in EXPLAIN output.

  12. Fix EXPLAIN (SETTINGS) to follow policy about when to print empty fields.

  13. Ensure plpgsql result tuples have the right composite type marking.

  14. Propagate sort instrumentation from workers back to leader.

  15. Make new regression test case parallel-safe, and improve its output.

  16. Push limit through subqueries to underlying sort, where possible.

  17. Fix inappropriate printing of never-measured times in EXPLAIN.

  18. Fix some infelicities in EXPLAIN output for parallel query plans.