Joshua D. Drake wrote:
> Heikki Linnakangas wrote:
>> Joshua D. Drake wrote:
>>> Heikki Linnakangas wrote:
>>>> Bruce Momjian wrote:
>>>>> o Automatic adjustment of bgwriter_lru_maxpages
>>>>>
>>>>> We show this as waiting for performance results. I am thinking we
>>>>> should hold this for 8.4.
>>>> Agreed. I spent close to a week trying different benchmarks and
>>>> configurations and simple test cases on a test server and my laptop, and
>>>> couldn't demonstrate bgwriter making a positive impact in any
>>>> configuration I tried. The theory behind the patch is sound, but it
>>>> looks like a lot more testing and analysis is needed.
>>> Wouldn't real world testing be needed to actually gain insight to this
>>> patch?
>> I would expect a fairly static benchmark workload to benefit from having
>> a bgwriter, more so than more unpredictable real world applications.
>
> Hmmm, I find that real world applications are quite predictable over
> time. Certainly you have spikes (good pr, whatever) but in general with
> a little bit of monitoring it is quite possible to evaluate a generally
> expected result.
>
> I guess my point is, if the patch looks good and does not appear to hurt
> anything, why not apply it? At least that way we can start to review the
> progress of the feature itself as it starts to see use.
If it doesn't appear to have any positive effect, why would we apply it?
If we apply the patch, how would you monitor it's effectiveness in a
live database? There's nothing to compare against.
-- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com