Re: PATCH: pgbench - random sampling of transaction written into log

On 30 Srpen 2012, 23:44, Robert Haas wrote:
> On Thu, Aug 30, 2012 at 3:48 PM, Tomas Vondra <tv(at)fuzzy(dot)cz> wrote:
>> That sounds like a pretty trivial patch. I've been thinking about yet
>> another option - histograms (regular or with exponential bins).
>
> I thought about that, too, but I think high-outliers is a lot more
> useful. At least for the kinds of things I worry about.

OK, let's fix the current patches first. We can add more options later.

>
>> What I'm not sure about is which of these options should be allowed at
>> the
>> same time - to me, combinations like 'sampling + aggregation' don't make
>> much sense. Maybe except 'latency-only-if-more-than + aggregation'.
>
> Maybe, but perhaps not even. I don't have a problem with saying that
> the user is allowed to pick at most one method of reducing the output
> volume. I think we could say: either sample, or take high outliers,
> or aggregate. If we want to make some of those work in combination,
> fine, but I don't think it's absolutely required. What WILL be
> required is a clear error message telling you what you did wrong if
> you use an unsupported combination.

I'll allow a single option - we can enable combinations that make sense
later.

>
> BTW, I think that using any of these options should automatically
> enable -l, rather than requiring it to be specified separately.

Good idea.

Tomas