Re: Allow a per-tablespace effective_io_concurrency setting - Mailing list pgsql-hackers

From Jeff Janes
Subject Re: Allow a per-tablespace effective_io_concurrency setting
Date
Msg-id CAMkU=1ytT-s37DMjx5GZpm1yNgDaXW=Gffe6LU_ZnD1=bWMQyg@mail.gmail.com
Whole thread Raw
In response to Re: Allow a per-tablespace effective_io_concurrency setting  (Josh Berkus <josh@agliodbs.com>)
List pgsql-hackers
On Wed, Sep 2, 2015 at 2:31 PM, Josh Berkus <josh@agliodbs.com> wrote:
On 09/02/2015 02:25 PM, Tomas Vondra wrote:
>
> As I explained, spindles have very little to do with it - you need
> multiple I/O requests per device, to get the benefit. Sure, the DBAs
> should know how many spindles they have and should be able to determine
> optimal IO depth. But we actually say this in the docs:

My experience with performance tuning is that values above 3 have no
real effect on how queries are executed.

Perhaps one reason is that the planner assumes it will get no benefit from this setting, meaning it is somewhat unlikely to choose the types of plans which would actually show a benefit from higher values.

Cheers,

Jeff

pgsql-hackers by date:

Previous
From: Merlin Moncure
Date:
Subject: Re: Allow a per-tablespace effective_io_concurrency setting
Next
From: Andres Freund
Date:
Subject: Re: Memory prefetching while sequentially fetching from SortTuple array, tuplestore