Horton

Re: [HACKERS] Runtime Partition Pruning

Alvaro Herrera <alvherre@alvh.no-ip.org>

View thread View on postgr.es
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: David Rowley <david.rowley@2ndquadrant.com>
Cc: Jesper Pedersen <jesper.pedersen@redhat.com>, Amit Langote <Langote_Amit_f8@lab.ntt.co.jp>, Rajkumar Raghuwanshi <rajkumar.raghuwanshi@enterprisedb.com>, Beena Emerson <memissemerson@gmail.com>, Robert Haas <robertmhaas@gmail.com>, amul sul <sulamul@gmail.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>, Dilip Kumar <dilipbalaut@gmail.com>
Date: 2018-03-31T15:41:32Z
Lists: pgsql-hackers 
David Rowley wrote:

> To put the new patch to the test, I tried pgbench -S -M prepared -s
> 100 with and without having modified pgbench_accounts to separate into
> 10 RANGE partitions of equal size.
> 
> A non-partitioned table was getting 12503 TPS.
> With partitioned tables, the old version of this patch was getting: 5470 TPS.
> With partitioned tables, the attached version gets 11247 TPS.
> For perspective, today's master with a partitioned table gets 4719 TPS.
> 
> So you can see it's a pretty good performance boost by skipping
> initialisation of the 9 non-matching subplans.  It's not hard to
> imagine the gains getting more significant with a larger number of
> partitions.

These are excellent news!

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Commits

  1. Initialize ExprStates once in run-time partition pruning

  2. Add bms_prev_member function

  3. Support partition pruning at execution time

  4. Document partprune.c a little better

  5. Fix crash in pg_replication_slot_advance

  6. Add parallel-aware hash joins.

  7. Fix typo.

  8. Allow --with-bonjour to work with non-macOS implementations of Bonjour.

  9. Mention need for --no-inc-recursive in rsync command