Love PerfStack..but next version can it have a brain of its own?

dodo123

The question we get asked a lot "on Server A between 6am-7am something happened and broke application z, can you see anything to show why?"

So my first thought is PerfStack so i go hunting through each element,component to see if anything has changed out of the ordinary.

Sometimes I get lucky most of the time the answer is "its probably in Solarwinds but the sheer amount of things to go through if you dont have a specific area to look you could be here all week."

So my idea is there a way to maybe use the baselines and just ask Perfstack "I know the period, I know the server what has changed in this period from maybe the same time last week"

Then you get a list of elements that have changed rather than everything.

Good idea or not?

jscheitel

I like the idea as an ad-hoc analysis tool as you describe.  The idea of recording baselines, ie "what is a normal pattern for this entity/server/application" is an interesting one anyway - but the hard thing obviously is "What is normal?" and "within what timeframe?" and to establish that, how much additional reference data do we need to keep on all those metrics we are not actively monitoring.  Daily, monthly? This would be some next level mojo if they could pull it off.