Server / server group statistics for hosting team
Hello,
Recently our company switched from new relic to Compuware
The hosting team asked me te create this kind of dashboard (new relic)
Each line is a server or server group with importent stats, example total request / min - memory - % errors, avarage cpu
So if i have 3 servers i like to see these stats:
Stats server 1, example total request / min - memory - % errors, avarage cpu
Stats server 2, example total request / min - memory - % errors, avarage cpu
Stats server 3, example total request / min - memory - % errors, avarage cpu
----------------------------------
Stats of group (1+2+3), example total request / min - memory - % errors, avarage cpu
The hosting team need a dashboard to project on a screen, green = ok, orange = warning, red = alarm 
----------------------------------
Extra question:
Is there a plugin to project dashboard on a wall screen?
Remark:
I tried to work with aplications buth these are fixed dashboards, example the response time = 0.4 ms,
we work wtih async webcall so i need the purepath duration (10 sec) to give 'response times'
Best Answer
Hi,
Exactly that dashboard won't be possible with the dashlets the dynaTrace client currently has. For example we don't have a control that shows a value as text and changes the color - but you could show the value as text and place a traffic light next to it that turns green/yellow/red. Depending on how you define the groups you might need to define business transactions to filter out the right values and group them as needed.
Extra question - dashboard on wall screen: Yes, the client has a fullscreen mode you can use for showing only dashboards. It is called kiosk mode - details here: Dashboards#KioskMode
Best, Roman
Thanks,
I have used the dynamic measure matrix to create dashboards,
I have placed a threshold on the average pure path duration,
Is it possible to place the threshold on the average time instead of the sample time?
in above example i do not expect an error because the average time is below the threshold.
I guess the percentile can be a solution? But still is it possible to generate alarm to averages?
Thx
Yes, but you will need to use incidents for doing that. You just defined a threshold on a measure (at least in the screenshot above), incidents are separate and define what should happen if this threshold is broken and also under which conditions (like your average over X minutes). In your example I would define the average response time over 5 minutes being above the threshold:
Check out the documentation for a good visual explanation of the evaluation timeframe: Incidents and Alerting
Best, Roman
If I am correct Compuware has profiles with groups (=one or more applications),
so i create a profile A, with a group B with multiple agent mappings (example webservices on server 1,2 &3)?
-------------------------------
Can i link information from the host agent into this groups?
I do not completely understand how to ty different information pieces together
