random_page_cost for tablespace

[Laszlo Nagy]

  Hi All,

We have a bigger table with some million rows. Number of index scans is
high, number of seq reads is low. This table if often joined with
others... so we want to buy a new SSD drive, create a tablespace on it
and put this big table on it. Random read speed on SSD is identical to
seq read. However, I need to tell the optimizer that random_page_cost is
less for the new tablespace. Is there a way to do it?

Thanks,

   Laszlo



--
Sent via pgsql-performance mailing list (pgsql-performance@...)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance

[Robert Haas]

2009/11/9 Laszlo Nagy <gandalf@...>:
> We have a bigger table with some million rows. Number of index scans is
> high, number of seq reads is low. This table if often joined with
> others... so we want to buy a new SSD drive, create a tablespace on it
> and put this big table on it. Random read speed on SSD is identical to
> seq read. However, I need to tell the optimizer that random_page_cost is
> less for the new tablespace. Is there a way to do it?

I happen to be working on a patch for this exact feature.  However,
even assuming it gets in, that means waiting for 8.5.

...Robert

--
Sent via pgsql-performance mailing list (pgsql-performance@...)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance

[Laszlo Nagy]

Robert Haas írta:

> 2009/11/9 Laszlo Nagy <gandalf@...>:
>  
>> We have a bigger table with some million rows. Number of index scans is
>> high, number of seq reads is low. This table if often joined with
>> others... so we want to buy a new SSD drive, create a tablespace on it
>> and put this big table on it. Random read speed on SSD is identical to
>> seq read. However, I need to tell the optimizer that random_page_cost is
>> less for the new tablespace. Is there a way to do it?
>>    
>
> I happen to be working on a patch for this exact feature.  However,
> even assuming it gets in, that means waiting for 8.5.
>  

That will be a very nice feature. Thank you! :-)


--
Sent via pgsql-performance mailing list (pgsql-performance@...)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance