Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From: =?utf-8?B?0J7Qu9C10LMg0KHQsNC80L7QudC70L7Qsg==?= <splarv@ya.ru>
Mime-Version: 1.0 (1.0)
Subject: Re: explanation for random_page_cost is outdated
Date: Fri, 24 Apr 2020 03:47:13 +0300
Message-Id: <72492300-BFD4-4CCF-9136-20CC41A18A10@ya.ru>
References: <20200319205646.GB28431@momjian.us>
Cc: pgsql-docs@lists.postgresql.org
In-Reply-To: <20200319205646.GB28431@momjian.us>
To: Bruce Momjian <bruce@momjian.us>
Precedence: bulk

Yep. Unclear. What parameter is recommended for SSD? Lower? 3? 2? 1?

Much better will be write: if you use SSD set 1.

=D0=9E=D0=BB=D0=B5=D0=B3

> 19 =D0=BC=D0=B0=D1=80=D1=82=D0=B0 2020 =D0=B3., =D0=B2 23:56, Bruce Momjia=
n <bruce@momjian.us> =D0=BD=D0=B0=D0=BF=D0=B8=D1=81=D0=B0=D0=BB(=D0=B0):
>=20
> =EF=BB=BFOn Thu, Feb 27, 2020 at 02:48:44PM +0000, PG Doc comments form wr=
ote:
>> The following documentation comment has been logged on the website:
>>=20
>> Page: https://www.postgresql.org/docs/12/runtime-config-query.html
>> Description:
>>=20
>> Explanation for random_page_cost is rather outdated, because it did only f=
or
>> case of mechanical hdd. But all modern database servers, which I know, ma=
de
>> upon SSD. Do or not do default value for random_page_cost equal to 1 is t=
he
>> question, but, IMHO, at list in the documentation  about random_page_cost=

>> need to add in a speculation about SSD.
>>=20
>> It's important because a business programming now is mostly web programmi=
ng.
>> Most database is poorly designed by web programmer, tables looked like a
>> primary key and a huge json (containing all) with large gin index upon it=
.
>> Now I am seeing a table with a GIN index 50% of the table size. The datab=
ase
>> is on SSD, of cause.  With default random_page_cost=3D4 GIN index don't u=
sed
>> by planner, but with random_page_cost=3D1 the result may be not excellent=
, but
>> acceptable for web programmers.
>=20
> Does this sentence in the random_page_cost docs unclear or not have enough=

> visibility:
>=20
>    https://www.postgresql.org/docs/12/runtime-config-query.html#RUNTIME-CO=
NFIG-QUERY-CONSTANTS
>   =20
>    Storage that has a low random read cost relative to sequential, e.g.
>    solid-state drives, might also be better modeled with a lower value for=

>    random_page_cost.
>=20
> --=20
>  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
>  EnterpriseDB                             https://enterprisedb.com
>=20
> + As you are, so once was I.  As I am, so you will be. +
> +                      Ancient Roman grave inscription +