DX Application Performance Management

Hi folks,

I have some applications being monitored and today I realized there were some differences on something that should be the same. I was comparing the metric Average Response Time for one Business Transaction and I realized the values for Average Response Time that appears under Business Segment for one agent was similar but not equal to Average Response Time for the same Business transaction but under the TESS Agent and then I have compared both with Average Response Time that is under by Business Service on the Triage Map tab and both show differences compared to the Triage Map metric. I got really confused with this information. So, what is the reason to have these differences for the information that should be the same? Somebody can help me to understand why these differences! (I included bellow some images showing the differences)

Metric under TESS Agent

Metric under Business Segment for an Agent

Metric under Triage Map (by Business Service)

Thank you.

Hi Carlos:

    Are all agents the same APM version? There have been various issues in earlier versions that may be impacting this.

Thanks

Hal German

Hi Hal,

Agent version is 9.5.5 and TESS Agent (that I presumed is the same version of EM) is 9.5.3.

The difference between these two categories of metrics is actually quite useful!

• Metrics from the TESS agent show performance as measured from the TIM / MTP (where the packet collection occurs via SPAN/Tap). This includes round trip time across the network between your collection point and your application server.
• Metrics from the Introscope agent show performance as measured from the JVM/CLR itself. This does not include any network round trip time. The Introscope agent re-uses the CEM definitions to analyze the handled HTTP requests and present the performance under the Business Segment node.

This means, when you see a large difference between the CEM / TESS metrics and the Introscope Agent metrics, you are seeing the performance impact of the network (switches, routers, etc.) and other infrastructure (web server, virtual switches, etc.). If the TESS metrics are indicating a very slow transaction, but the Introscope agent metrics are not (for that same Business Transaction / Business Segment definition), then the performance problem is due to environment, not the application itself.

The Application Triage Map attempts to bring both categories together in one view to help perform this comparison.

Hi Jakbutler, thanks for this explanation.

So, you are telling me the metric under Business Segment is originated by the Introscope Agent and it shows the agent point of view for the transaction, right? The same way the one under TESS agent reflects the measurement done based on TIM data... ok.

But why none of those are equal to the values displayed on the Triage Map? Where this measurement came from? Is these values "adapted" from any metric shown on the Metric Browser tree?

Hi Carlos,

There are two types of metrics reported under the Business Service node in Triage map, Customer experience metrics (comes from Tess Agent) and BT metrics that comes from Agent.

Agent metrics are aggregated i.e if you have N JVM's monitoring same Business Transaction then metrics reported will be aggregated depending on the metric type(Avg. or Per Interval) and at the bottom it will show how many location\agents were matched for same BT. So check if there are multiple Agents, if not then make sure the resolution used for viewing Triage Map and Agent data is same.

If they still don't match then open a support issue so we can investigate further.

Regards,

Kulbir.