Re: Let PostgreSQL's On Schedule checkpoint write buffer smooth spread cycle by tuning IsCheckpointOnSchedule?

On 07/05/2015 08:19 AM, Fabien COELHO wrote:
>> I am a bit skeptical about this. We need test scenarios that clearly
>> show the benefit of having and of not having this behavior. It might be
>> that doing this always is fine for everyone.
>
> Do you mean I have to proove that there is an actual problem induced from
> this patch?

You don't have to do anything if you don't want to. I said myself that
this needs performance testing of the worst-case scenario, one where we
would expect this to perform worse than without the patch. Then we can
look at how bad that effect is, and decide if that's acceptable.

That said, if you could do that testing, that would be great! I'm not
planning to spend much time on this myself, and it would take me a fair
amount of time to set up the hardware and tools to test this. I was
hoping Digoal would have the time to do that, since he started this
thread, or someone else that has a system ready for this kind of
testing. If no-one steps up to the plate to test this more, however,
we'll have to just forget about this.

> Having a guc would also help to test the feature with different values
> than 1.5, which really seems harmful from a math point of view. I'm not
> sure at all that a power formula is the right approach.

Yeah, a GUC would be helpful in testing this. I'm hoping that we would
come up with a reasonable formula that would work well enough for
everyone that we wouldn't need to have a GUC in the final patch, though.

- Heikki