Right, the use case Jeroen presented to me was things where ever so
slightly stale statistics on prod cause bad behaviour, which we can't
then reproduce later: so we should consistently and pervasively gather
this data to counter that.
I do have a concern about perf impact on prod doing this, would like
Stuarts input there.
Right, the use case Jeroen presented to me was things where ever so
slightly stale statistics on prod cause bad behaviour, which we can't
then reproduce later: so we should consistently and pervasively gather
this data to counter that.
I do have a concern about perf impact on prod doing this, would like
Stuarts input there.