Re: Hardware suggestions for high performance 8.3

От: Henrik
Тема: Re: Hardware suggestions for high performance 8.3
Дата: ,
Msg-id: 1D62209E-DFEE-4994-8B5B-BC181608B0AE@mac.se
(см: обсуждение, исходный текст)
Ответ на: Re: Hardware suggestions for high performance 8.3  (Greg Smith)
Список: pgsql-performance

Скрыть дерево обсуждения

Hardware suggestions for high performance 8.3  (Henrik, )
 Re: Hardware suggestions for high performance 8.3  ("Claus Guttesen", )
  Re: Hardware suggestions for high performance 8.3  (Henrik, )
 Re: Hardware suggestions for high performance 8.3  (Matthew Wakeling, )
  Re: Hardware suggestions for high performance 8.3  (Henrik, )
   Re: Hardware suggestions for high performance 8.3  ("Claus Guttesen", )
   Re: Hardware suggestions for high performance 8.3  (Matthew Wakeling, )
   Re: Hardware suggestions for high performance 8.3  (Greg Smith, )
    Re: Hardware suggestions for high performance 8.3  (Henrik, )
    Re: Hardware suggestions for high performance 8.3  (Henrik, )
     Re: Hardware suggestions for high performance 8.3  (Greg Smith, )
      Re: Hardware suggestions for high performance 8.3  ("Scott Marlowe", )
      Re: Hardware suggestions for high performance 8.3  ("Joshua D. Drake", )
 Re: Hardware suggestions for high performance 8.3  (Greg Smith, )

25 jun 2008 kl. 17.45 skrev Greg Smith:

> On Wed, 25 Jun 2008, Henrik wrote:
>
>> Would you turn off fsync if you had a controller with BBU? =)
>
> Turning off fsync has some potential to introduce problems even in
> that environment, so better not to do that.  The issue is that you
> might have, say, 1GB of OS-level cache but 256MB of BBU cache, and
> if you turn fsync off it won't force the OS cache out to the
> controller when it's supposed to and that can cause corruption.
>
> Also, if you've got a controller with BBU, the overhead of fsync for
> regular writes is low enough that you don't really need to turn it
> off. If writes are cached the fsync is almost free.
Thanks for a thoroughly answer. I guess I wont be turning of fsync. :)

Thanks Greg!



>
>
> --
> * Greg Smith  http://www.gregsmith.com
> Baltimore, MD
>
> --
> Sent via pgsql-performance mailing list (
> )
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-performance



В списке pgsql-performance по дате сообщения:

От: Craig James
Дата:
Сообщение: Typecast bug?
От: Craig James
Дата:
Сообщение: Re: RAID 10 Benchmark with different I/O schedulers