Dear All, @Kris Ziemianowicz
We wanted to alert on the # of active sessions but the metric is not available for the alerting even though we used in the DMI report.
e.g. below metrics are not available for the Monitored metric in the Triggering and propagation settings and maybe many more can be discover later on.
Does anyone has faced the same situation?
What can be the solution to this situaiton?
Release 2018 has a new alerting system, closely integrated with DMI, it will make it easier to add all DMI metrics. Did you have a look at 2018 beta?
Anyway, when I see the metrics you mention, I have to ask a question on the use case your are after? Please note that the Active Sessions metric works for HTTP analyzer only, it tells how many analyzed web sessions are open within AMD. It doesn't tell how many TCP sessions are in progress.
So is your case about open TCP sessions? And if so - how do you define number of open TCP sessions? Strictly parallel or a counter of opened sessions within a monitoring interval? Or open (not opened) sessions within an interval, or something else? And most importantly - what should this counter represent?
OK, if the session is an HTTP session, then this makes sense. Alternatively the counter of unique users could be helpful. These would give different results when aggregated over the day:
if I'm the only user active on a web site, for say 30 min during the day, then counter of sessions will be 1 in every interval and it will be 30 for the day if monitoring interval is 1-minute. But number of unique users will be 1 regardless of reporting period chosen.
I just want clarify:
It is true that number of dimensions and metrics available for alerting in 2018 release is significantly increased. However it doesn't mean that all DMI metrics and dimensions are available for alerting.
Some of them do not make sense for alerting (i.e. dimension Time), some of them are not available because we consider them as a rarely used even on reports. But, as Kris already mentioned, with new DMI based mechanism it will be much easier to add new metrics to alerting mechanism if there is such need.