none
Abnormal behavior Monitoring RRS feed

  • Question

  • Hi

            Its good to play with Application insights and query related stuff. I am wondering how the Application insights reporting abnormal behavior. like i have scenario:

    I can query which requests taking long time. But i am unable to see which requests are exceeding from their normal behavior and how many times in a day they are exceeding. Support ,we have thousands of requests in a day and i want to see how may requests are taking long and how many time they are being exceeded from their normal behavior. Any Hint or Direction:I also want to create alerts in this scenario.

    like , i want to see what is the failure rate or rate of going out of mentioned threshold of requests in my system. or any kind of comparative analysis.

    Thanks!




    • Edited by IbrahimUmar Wednesday, December 19, 2018 11:28 AM Modify Q
    Wednesday, December 19, 2018 11:19 AM

All replies

  • Hi IbrahimUmar,

    Thanks for using Application Insights.

     1. For identifying and alerting Abnormal behavior for your application with in Application Insights –

    • Please check the feature called as “Smart Detection” with in Application Insights , which have some predefined set of smart detection rules and you also can get alerted via email when the rule gets trigged.

    • Current list of Smart Detection Rules 

    2. You can also look for “Performance” feature with in Application Insights , which gives you an over view of Request Count, Time Taken based on Operation Names and also Insights for a specific operation.

    3. You can always pin these graphs to your Azure Dashboard and also “View in Analytics” and create a alert based on your requirement. 

    Additional References

    https://docs.microsoft.com/en-us/azure/application-insights/app-insights-proactive-diagnostics https://docs.microsoft.com/en-us/azure/application-insights/app-insights-proactive-performance-diagnostics https://docs.microsoft.com/en-us/azure/application-insights/app-insights-proactive-failure-diagnostics

    https://docs.microsoft.com/en-us/azure/application-insights/app-insights-tutorial-alert

    Hope the above information helps you to further dive into Application Insights features. Please let us know if you have any further questions. Thanks

    Wednesday, December 19, 2018 4:40 PM
    Owner
  • Thanks!!

    I think you couldn't get me. Like , i have a query:

    requests
    | where timestamp > ago(12h) and duration > 1000
    | summarize sum(itemCount) by url, bin(timestamp, 1h), performanceBucket, success,duration
    | sort by performanceBucket desc

    | render timechart

    I can start getting info/alerts that which requests are long. This is ok.

    Required: I want to see from last 2-3 days or a week , which requests were working find but suddenly they start showing abnormal behavior. This information should be showed by history. Like , apart from all long running or time taking request , i can report these are important to see request which never been taken more than half of the second but now showing this. We need to check this WHY.


    Hope i explained this.

    Thank You


    WSBukhari

    Thursday, December 20, 2018 2:28 PM
  • Modifying your query to include items with a timestamp that goes back to your preferred time (3d or 7d) should give you the result you're looking for.  You can modify which data points you are interested in, and further investigate events of interest:

    requests
    | where timestamp > ago(7d)
    and duration > 
    1000 
    | summarize sum(itemCount) 
    by url, bin(timestamp, 1h), performanceBucket, success,duration
    | sort by performanceBucket desc
    | render timechart

    Thursday, January 3, 2019 1:18 AM
    Moderator
  • But how this query showing me the abnormality of particular request? or if it is showing , may be i am unable to understand you. Kindly elaborate

    WSBukhari

    Thursday, January 3, 2019 11:24 AM
  • The abnormality is what you define in your query.  In this case, it's a request with a long duration.  You could just as easily look for high request rates, failures, page views, load times, and much more.
    Friday, January 4, 2019 9:35 PM
    Moderator