Re: increasing shared buffers: how much should be removed from OS filesystem cache?

On 4-Sep-06, at 8:07 AM, Guillaume Cottenceau wrote:

> Dave Cramer <pg 'at' fastcrypt.com> writes:
>
>> Guillaume
>>
>> 1G is really not a significant amount of memory these days,
>
> Yeah though we have 2G or 4G of RAM in our servers (and not only
> postgres running on it).
>
>> That said 6-10% of available memory should be given to an 8.0 or
>> older version of postgresql
>>
>> Newer versions work better around 25%
>>
>> I'm not sure what you mean by mechanically removed from
>> effective_cache
>
> I mean that when you allocate more memory to applications, the
> consequence is less memory the OS will be able to use for disk
> cache.
>
>> effective cache is really a representation of shared buffers plus OS
>> cache
>
> Are you sure the shared buffers should be counted in? As I
> understand the documentation, they should not (as shared buffers
> is allocated memory for the OS, not part of "kernel's disk
> cache"):
Yes, I am sure this should be counted, however effective_cache is not
actually allocating anything so it doesn't have to be exact, but it
has to be in the correct order of magnitude
>
>     Sets the planner's assumption about the effective size of the
>     disk cache (that is, the portion of the kernel's disk cache
>     that will be used for PostgreSQL data files). This is
>     measured in disk pages, which are normally 8192 bytes each.
>     The default is 1000.
>
> --
> Guillaume Cottenceau
> Create your personal SMS or WAP Service - visit http://
> mobilefriends.ch/
>