Re: Enable data checksums by default

From: Bernd Helmle <mailings(at)oopsware(dot)de>
To: Christoph Berg <myon(at)debian(dot)org>, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Enable data checksums by default
Date: 2019-03-29 10:16:11
Message-ID: c5d3e07dcb09d4f3f52e0eea691268cfb7f9cc71.camel@oopsware.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Am Dienstag, den 26.03.2019, 16:14 +0100 schrieb Christoph Berg:
> select 92551.0/97363;
> 0.9506
>
> So the cost is 5% in this very contrived case. In almost any other
> setting, the cost would be lower, I'd think.

Well, my machine (Intel(R) Core(TM) i7-6770HQ CPU @ 2.60GHz, 32 GByte
RAM) tells me this:

pgbench -s 50 -i pgbench
pg_ctl -o "--shared-buffers=128kB" restart
pgbench -r -P4 -Mprepared -T60 -c $clients -j $clients -n -S

...prewarm...

Clients checksums
1 20110
2 35338
4 67207
8 96627
16 110091

Clients no checksums
1 21716
2 38543
4 72118
8 117545
16 121415

Clients Impact
1 0,926045312212194
2 0,916846119918014
4 0,931903269641421
8 0,822042621974563
16 0,906733105464728

So between ~7% to 18% impact with checksums in this specific case here.

Bernd

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message David Steele 2019-03-29 10:27:11 Re: Add exclusive backup deprecation notes to documentation
Previous Message Adrien NAYRAT 2019-03-29 10:09:46 Re: Log a sample of transactions