Re: Tuplesort merge pre-reading

From: Gavin Flower <GavinFlower(at)archidevsys(dot)co(dot)nz>
To: Peter Geoghegan <pg(at)heroku(dot)com>
Cc: pgsql-hackers(at)postgresql(dot)org, Claudio Freire <klaussfreire(at)gmail(dot)com>
Subject: Re: Tuplesort merge pre-reading
Date: 2016-09-12 00:34:48
Message-ID: 6f410b28-d16e-732c-7dff-6486d210ef7d@archidevsys.co.nz
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On 12/09/16 12:16, Gavin Flower wrote:
[...]
> two blocks would be logically adjacent (which means they are likely
> to be physically close together, but not guaranteed!).
>
[...]

Actual disk layouts are quite complicated, the above is an over
simplification, but the message is still valid.

There are various tricks of disc layout ( & low level handling) that can
be used to minimise the time taken to read 2 blocks that are logically
adjacent. I had to know this stuff for discs that MainFrame computers
used in the 1980's - modern disk systems are way more complicated, but
the conclusions are still valid.

I am extremely glad that I no longer have to concern myself with
understanding the precise low stuff on discs these days - there is no
longer a one-to-one correspondence of what the O/S thinks is a disk
block, with how the data is physically recorded on the disc.

Cheers,
Gavin

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Fabrízio de Royes Mello 2016-09-12 00:51:13 CommitFest 2016-09 status summary
Previous Message Gavin Flower 2016-09-12 00:16:23 Re: Tuplesort merge pre-reading