Re: Number of Shared Blocks Hit

From: Andres Freund <andres(at)anarazel(dot)de>
To: luis(dot)roberto(at)siscobra(dot)com(dot)br
Cc: pgsql-bugs <pgsql-bugs(at)lists(dot)postgresql(dot)org>
Subject: Re: Number of Shared Blocks Hit
Date: 2021-04-02 17:35:09
Message-ID: 20210402173509.t37bxjrnbjdsop2r@alap3.anarazel.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

Hi,

On 2021-03-04 08:23:07 -0300, luis(dot)roberto(at)siscobra(dot)com(dot)br wrote:
> While running a update, and checking EXPLAIN ANALYZE output, I found
> it strange that in the "ModifyTable" node, it shows "Shared Hit
> Blocks":351938580. If my math is correct, that amounts to more than
> 2.5TB.

Note that shared blocks hit can include repeated hits to the same
buffer. Over and over again. E.g. when you insert a new index row for
each of the rows, parts of the index will have to be traversed for each
row. Those pages will all be in the cache, hence no increased "Shared
Read Blocks", but will be counted as separate hits.

Remembering which pages we accessed previously would be quite
expensive...

Greetings,

Andres Freund

In response to

Browse pgsql-bugs by date

  From Date Subject
Next Message PG Bug reporting form 2021-04-03 23:07:25 BUG #16952: PG admin errors out when connecting via ssh because it can't get pg_settings
Previous Message Boris P. Korzun 2021-04-02 14:09:14 Re: Inconsistent behavior of pg_dump/pg_restore on DEFAULT PRIVILEGES