3 of 3 people found this helpful
Agent Thresholds - These are pconfig settings sent down to the PatrolAgent (like PCM). Setting a threshold like this will require PATROL_EV propagation and message rewording of the PATROL_EV standard event. THe PA will workout the threshold breach and send an event.
Servers Threshold - These are thresholds set on the TSIM. Data from the PA is streamed up and the threshold breach is worked out by the analytics engine sending an "Intelligent" ALARM event to the cell
1 of 1 people found this helpful
as correctly explained in short by Gordon James the agent thresholds are the same thresholds as you used with patrol agents from the beginning, its static thresholds and they are defined on agent level
the server thresholds use the data in TSIM DB and the big advantage is that you have all the options TSIM is providing, including baselining of performance values, predictive events and so on
2 of 2 people found this helpful
In addition to what Gordon and Roland very well explained, one of the major differences between agent threshold and server threshold is who is responsible to generate events when thresholds are violated. For agent thresholds, PATROL agent is responsible to generate events as the thresholds are pushed down from CMA/TSPS and stored in each PATROL agent. For server thresholds, BPPM server/TSIM is responsible to generate events since the thresholds are stored in BPPM server/TSIM only and PATROL agent is not aware of these thresholds.
If you use agent thresholds, make sure your filter policy is set as 'Send events only' or 'Send data and events'. Otherwise the events generated from PATROL agents won't be displayed in BPPM/TSPS console.
If you use server thresholds, make sure your filter policy is set as 'Send data only' or 'Send data and events'. Otherwise the data collected by PATROL agents won't be sent to BPPM server/TSIM. BPPM server/TSIM needs these data to see if they are violating the server thresholds.
There are other differences such as server thresholds are good for baselines and agent thresholds are good for recovery actions, etc. But you will need to know the monitoring requirements first before deciding which one is better for you. And it is perfectly OK to use server thresholds for one KM and agent thresholds for another KM.
Hope it helps,