Re: x206-x225

From: PFC <lists(at)peufeu(dot)com>
To: "Joost Kraaijeveld" <J(dot)Kraaijeveld(at)askesis(dot)nl>, "David Lang" <david(at)lang(dot)hm>
Cc: "Richard Huxton" <dev(at)archonet(dot)com>, pgsql-performance(at)postgresql(dot)org
Subject: Re: x206-x225
Date: 2006-03-11 11:33:50
Message-ID: op.s58yeob1cigqcu@apollo13
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance


>> each transaction requires a sync to the disk, a sync requires a real
>> write (which you then wait for), so you can only do one transaction per
>> rotation.
> Not according to a conversation I had with Western Digital about the

It depends if you consider that "written to the disk" means "data is
somewhere between the OS cache and the platter" or "data is writter on the
platter and will survive a power loss".

Postgres wants the second option, of course.

For that, the data has to be on the disk. Thus, the disk has to seek, wait
till the desired sector arrives in front of the head, write, and tell the
OS it's done. Your disk just stores data in its embedded RAM buffer and
tells the OS it's written, but if you lose power, you lose anything that's
in the disk embedded RAM cache...

Advanced RAID cards have battery backed up RAM cache precisely for that
purpose. Your harddisk doesn't.

In response to

Responses

Browse pgsql-performance by date

  From Date Subject
Next Message Joost Kraaijeveld 2006-03-11 14:26:14 Re: x206-x225
Previous Message Joost Kraaijeveld 2006-03-11 08:17:09 Re: x206-x225