From: | Gavin Flower <GavinFlower(at)archidevsys(dot)co(dot)nz> |
---|---|
To: | Marc Mamin <M(dot)Mamin(at)intershop(dot)de>, "'KONDO Mitsumasa'" <kondo(dot)mitsumasa(at)lab(dot)ntt(dot)co(dot)jp>, Stephen Frost <sfrost(at)snowman(dot)net>, Dimitri Fontaine <dimitri(at)2ndQuadrant(dot)fr> |
Cc: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Andrew Dunstan <andrew(at)dunslane(dot)net>, Peter Geoghegan <pg(at)heroku(dot)com>, Robert Haas <robertmhaas(at)gmail(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Add min and max execute statement time in pg_stat_statement |
Date: | 2013-10-23 21:57:18 |
Message-ID: | 5268463E.4010806@archidevsys.co.nz |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On 24/10/13 10:34, Marc Mamin wrote:
>>> Oscillating plan changes may fit multimodal but I don't feel that's
>>> typical. My experience has been it's either an extremely rare plan
>>> difference or it's a shift from one plan to another over time.
>> After all, all of avg, min, max and stdev are only numerical value for predicting model. There aren't the robustness and strictness such as Write Ahead Logging. It resembles a weather forecast. They are still better than nothing.
>> It is needed a human judgment to finally suppose a cause from the numerical values. By the way, we can guess probability of the value from stdev.
>> Therefore we can guess easily even if there is an extreme value in min/max whether it is normal or not.
>>>> What I've been gathering from my quick chat this morning is that
>>>> either you know how to characterize the distribution and then the min
>>>> max and average are useful on their own, or you need to keep track of
>>>> an histogram where all the bins are of the same size to be able to
>>>> learn what the distribution actually is.
> Hello,
>
> We have an in house reporting application doing a lot of response times graphing.
> Our experience has shown that in many cases of interest (the one you want to dig in)
> a logarithmic scale for histogram bins result in a better visualization.
> attached an example from a problematic postgres query...
>
> my 2 pences,
>
> Marc Mamin
Looks definitely bimodal in the log version, very clear!
Yes, I feel that having a 32 log binary binned histogram (as Alvaro
Herrera suggested) would be very useful. Especially if the size of the
first bin can be set - as some people would like to be 100us and others
might prefer 1ms or something else.
Cheers,
Gavin
From | Date | Subject | |
---|---|---|---|
Next Message | Peter Geoghegan | 2013-10-23 22:26:09 | Re: Add min and max execute statement time in pg_stat_statement |
Previous Message | Josh Berkus | 2013-10-23 21:54:01 | Autotuning and RAM survey results |