Azure Application Insights
Azure Application Insights is an extensible Application Performance Management (APM) service designed for Developers, DevOps, and Site Reliability Engineering (SRE) professionals. It is a unique, built-in Native feature of Azure under Azure Monitor within the Azure ecosystem and is used to monitor live applications. Its main job is to automatically detect performance issues such as failed requests, response times, and application availability. It also includes several Investigative methods for Diagnosing, troubleshooting anomalies and runtime behaviour of the application.
It supports applications built using various platforms, including.NET, Node.js, Java, and Python, regardless of whether they are hosted on-premises, in a hybrid environment, or on any public cloud. And it will monitor both Frontend and Backend services.
You can mointor realtime insights of the Application behavior in a dedicated Azure Application Insights Dashboard for your application as shown in below snapshot.
Veerpedia Azure App Insights Dashboard
Note: Configuration and Instrumentation of Azure Application Insights will be published soon in upcoming article, here we will discuss in APM and Observability Perspective.
Investigation Methods in Azure Application Insights
There some Investigative features availble in Azure Application Insights for troubleshooting and monitoring real time behaviour of the application, The following features are:
- Application Dashboard
- Application Map
- Live metrics
- Search View
- Avilability View
- Failures View
- Performance View
- Agent Details
These all features you can able to navigate in the Azure Portal through following path Azure Portal → Resource Manager | Resource Groups → Your Resource Group → Application Insights → Investigate. Please refer below attached snapshot for your understanding.
Application Dashboard
In Application Dashboard we can see complete behavior and realtime stats of your application, and followed by we can also able to monitor Key Performance Indicators based on Appplication Performance such as Usage, Reliability, Availability, Server Response Time, Number of Sessions and User count, Failed requests, Average Page Loadtime and other important KPI's.
We can also create a customize application dashboard as per our feasability and requirment, and followed by we can also do these:
- Download the Exisiting Dashboard in the form of JSON and also take a print
- Clone the Dashboard to make changes
- Assign tags and Add filters
- We can also share dashboard in between teams and can restict also
- Track Dashboard modifications under Manage History
Please refer below attached snapshot for your understanding on Application Dashboard
Application Map
Application Map will give you topology representation of your complete application, and it automatically found the relationships between the application internal components and their dependencies. For suppose if a user trying to access your web application it will shows request flows, database calls, request latency and Azure Cache for Redis. The Application Map will display these components as Connected nodes along with their health status and dataflow. Please refer below attached snapshot for your understanding on Application Map.
Live metrics
Live metrics is the Key feature of Azure Application Investigation, it will provides realtime monioring of your application, Developers frequently use this feature during Software Release deployments and troubleshooting production releated issues. We can see these live metrics such as:
- Incoming Request Rate
- Reponse Time of the Request
- Live request details
- CPU and Memory usage
- Exception Count
- Dependency calls for Databases and app services (.NET, Springboot and Python)
Please refer below attached snapshot for your understanding on Live Metrics
Search View
Search view in Azure Application Insights is used to explore individual requests and used to find out the failure execption. It will give a glance on request trace which means complete request journey, Search view will give these request parameters such as:
- Timestamp
- Transaction ID or Trace ID
- Failure status with Error Message
- Exception message and Stack Trace
- Logs for Specific Failure Transaction
- Request Correlation ID
To perfrom complete deep dive into each failure request, Search view will help us to find out the root cause analysis also, and we can also share the respective search view analysis accross team members by share the Search View URL. Please refere below attached snapshot for your understanding on Search View.
Availability View
Availability view will tell us whether the end user can access the application and respond as expected or not by performing URL pings and Web Tests to validate response status, and it can also check the SSL certificate validity to ensure a secure connection between the user and the application.
This availability view helps us identify downtime and any network connectivity issues occurring. And Developers, SRE, and DevOps engineers can configure alerts to notify respective teams when the tests have failed. Mostly as Standard test can be performed from various Azure Geolocations.
For example, you are the owner of any e-commerce application and using an Availability test that verifies the login page every 5 minutes from multiple locations. If the website returns a 503 Service Unavailable error, the Test View will show the failed test locations and outage duration, so that this can help the Operations team/ Developers to act on them immediately.
Please refer below attached snapshot for your understanding on Availbility View.
Failures view
The failure view will give a big picture of failure trends regarding your application errors by demontrating failed requests, exceptions, and dependency failures. It points out the most occurring errors, failures trends, and operations that are effected. This will enable DEV/SRE/Operations teams to lookup and identify the cause of issues and regulate application downtime without much impacting the business transactions.
Note: Before you start looking into Failure view, please take a look at the Failed requests graph by hovering the cursor, so that you can understand the present normal-day failure trend and compare it with the trend that has occurred during Software release deployments, as shown in the above attached snapshot.
Performance View
The performance view will tell us how efficiently your application is perfroming by metrics such as response time, request duration, throughput and latency for dependecicies. It will points out slow performing requests and identify perfromance bottlenecks that may show impact on the user experince. The performance of the application mostly depends on the server response time and server requests.
So before directly deep dive into Performance view, I would suggest you to observe the Respose time trend by Server requests. Historical trends on performance will help monitoring thr application perfromance over time and validate improvements after Software release deployements.
For example, customers are complaining that the article search page is slow, and the performance shows that the GET/search API response time graph spikes from 400 ms to 3 seconds because an external article catalog API is responding very slowly. This will help and give insight to the development team to optimize the dependency call, which can reduce the response time.
Agent Details (Preview)
Agent preview is used to verify the monitoring agent or SDK that is collecting the telemetry data from the application. It includes the runtime environment, connected instances, and status of telemetry collection. This feature is very much useful for validating that instrumentation is configured correctly and any missing data or incomplete data.
For example, a newly deployed backend Spring boot service is not appearing in Azure Application Insights. And the Agent preview shows that the application is using an Outdated SDK versions, and telemetry data is not being sent correctly. After the developer team updating the Azure Application Insights SDK and connection string, the flow of telemetry data started, and we were also able to see the backend Spring Boot service in the Application Dashboard. Please refer below attached snapshot for your undersatnding on Agent Preview.
