Shared Instance Services and Response Time Alerts

[NickN]

How are you all handling the response time alerts with regards to the services that are on the shared pool? It seems that 98% of the alerts I get are from these when they're hit for the first time each day. Do you just shut the alerts off for them or are you using some other alert configuration than the out of the box one?

[AndrewSakowicz]

What is  the performance of the "first time hit" vs. the consecutive?  Are these "first time hit" alerts valid? Is the challenge there are many alerts in the UI ?   

[NickN]

For example I have a service that will have a max response time of 11 seconds at what appears to be the first "hit" of the day, then all the subsequent max responses are below a second.

I'm just curious how others handle this, or what best practice is? I deployed the new Monitor shortly after it was released. We have 220 services deployed and since I turned on Monitor it's generated 12,721 warnings and 10,804 criticals...98% of these are from the max response time when a pooled service is just waking up.

Here's a graph:

Spikes first thing in the morning, triggering the warning, then it's fine throughout the day.

[AndrewSakowicz]

Thank you for the info.  Very helpful.  It appears, this condition impacts min 3 requests and lasts approximately 15 min, (3 points above, with 5 min interval).  If you change the alert sample from 3 to 5 (and potentially change aggregation to p95) the alerts will not fire for the above.