How can we extract Baselines & Standard Deviation metrics out of AppDynamics (On-Premise) ?

[Anshuman.Bhatia]

I am looking to extract the Response Times, Baselines and Std. Deviation metrics out of our AppDynamics controller for analyzing the Business Transactions.

Now I can extract the response times for the BTs , but I am still unable to find any way to extract the baselines and standard deviations anywhere using Metric Browser or the AppD REST API documentation.

Is there any way that I can use for extracting these specific metrics ? And if so, can you please point me to a link that details the steps ?

-Anshuman

[alakshya.maheria]

Hi,

Open Metrics Browser, select a business transaction from business transaction performance. Add Average Response Time (ms). you can select baseline option to see the baseline.

Regards,

[alakshya.maheria]

You can refer to https://community.appdynamics.com/t5/Knowledge-Base/Plot-Standard-Deviation-for-Business-Transaction...

Regards,

[Anshuman.Bhatia]

Thanks for replying Alakshya. I rather wanted to query these values via REST API to analyze these on an EXTERNAL system. 

I can now see that my question wasn't very clear and I apologize for that. 

[alakshya.maheria]

Hi,

2 Options that I see.

a) if you wish to have it displayed as a dashboard, then you can create a time series widget. The data should be Busines Transaction performance. Select metric as average response time and select include baseline.

b) If need to export into some file, then you will have to create a script which does 2 things

    Use the configuration api to get the business transactions

    /controller/rest/applications/<application id>/business-transactions

   Get the BT names from the xml.

   Use metrics and snapshot api https://docs.appdynamics.com/display/PRO43/Metric+and+Snapshot+API and get the business transaction metrics based on the above data.

I have not tried this, but I think it should be the approach.

Regards,

[Anshuman.Bhatia]

Hello,

Before posting the question , I did try out the second option and this does NOT give me the baseline and standard deviation values. I have pasted below the exact response that I get when looking at metric data for Past 5 Mins using the metric snapshot API -

https://appdynamicsInstance.com/controller/rest/applications/SeinfeldRocks/metric-data?metric-path=O...

<metric-datas><metric-data>
  <metricId>80761</metricId>
  <metricPath>Overall Application Performance|Average Response Time (ms)</metricPath>
  <metricName>BTM|Application Summary|Average Response Time (ms)</metricName>
  <frequency>ONE_MIN</frequency>
  <metricValues>
    <metric-value>
      <startTimeInMillis>1513161900000</startTimeInMillis>
      <value>1324</value>
      <min>0</min>
      <max>270200</max>
      <current>40</current>
      <sum>3225950</sum>
      <count>2437</count>
      <standardDeviation>0.0</standardDeviation>
      <occurrences>0</occurrences>
      <useRange>true</useRange>
    </metric-value>
  </metricValues>
</metric-data>
</metric-datas>

No matter what metric I use (GC Time , BT Response Time , BT Calls per Min, Heap Usage, CPU Usage), the Standard Deviation and Occurrences are always 0 while obviously the Metric Browser does shows the values. For the metric response pasted above , the occurrences, baseline and standard deviation values are shown as 30 , 4926 ms and 46000 ms respectively on the metric browser. 

The first option is irrelevant as I need to analyze the data outside of AppDynamics and hence the time widget is not what I need.

Thanks for your help !

[alakshya.maheria]

The dynamic baselining is calculated based on the baseline condition that are defined. It could be last 30 days, last 60 days or what is setup. 

I am not sure how these values can be pulled out using an API, but given the understanding that baseline is what defines what is normal for a metric, you could try to pull out the Normal Average Response Time (ms) metric using Rest API and compare it with average response time. I am not sure how far the data might match with what baseline is in the system but it could be an option for representation.

Regards,