Explain analyse with track_io_timing

From: Jayadevan M <maymala(dot)jayadevan(at)gmail(dot)com>
To: pgsql-performance(at)lists(dot)postgresql(dot)org
Subject: Explain analyse with track_io_timing
Date: 2022-03-10 05:10:17
Message-ID: CAFS1N4j=QMEsQgnb4ZLjoAgZNwaK-aNkL8y9EBvFVV2PS=dZ8w@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

Hello team,

What is the unit of I/O Timings in explain (analyze, buffers) ? There is a
plan with quite a few nodes. In each case, the value of I/O Timings is
much more than the time for the outer node. A few lines from the plan -

Hash Left Join (cost=14320945.22..7099974624.27 rows=194335062701
width=5511) (actual time=107913.021..*108109*.313 rows=759 loops=1)
Buffers: shared hit=738871 read=1549646, temp read=92710 written=92973
I/O Timings: read=*228324*.357
-> Hash Left Join (cost=14049069.69..246411189.41 rows=18342148438
width=5467) (actual time=96579.630..*96774*.534 rows=759 loops=1)
Buffers: shared hit=684314 read=1377851, temp read=92710
written=92973
I/O Timings: read=*217899*.233
At the end, there is
Execution Time: 108117.006 ms

So it takes about 108 seconds. But the I/O Timings are higher.

Best Regards,
Jay

Responses

Browse pgsql-performance by date

  From Date Subject
Next Message Julien Rouhaud 2022-03-10 06:05:28 Re: Explain analyse with track_io_timing
Previous Message aditya desai 2022-03-09 06:09:55 Re: Any way to speed up INSERT INTO