Today we are excited to announce the Public Preview of Azure API Management (APIM) integration with Azure Application Insights (AI)! This new feature lets customers add APIM telemetry to AI and use AI’s rich set of capabilities to monitor and troubleshoot their APIs.
How do I enable the integration?
- Go to Azure Portal and navigate to your APIM instance
- Click on Application Insights menu (located near the bottom)
- Check Enabled
- Click on Select Application Insights and select the AI instance you want to associate with this APIM service
- You are all set!
What data is added to AI?
Is there any performance impact?
Yes, there is, and its magnitude depends on incoming request rate. Based on our load tests, after enabling this feature, we observed a 40%-50% reduction in throughput when request rate exceeded 1,000 requests per second. AI is designed to use statistical analysis for assessing application performances. It is not intended to be an audit system thus is not suited for logging each individual request for high-volume APIs, as we are doing now. We will add sampling controls before the GA to help customers reduce telemetry traffic and storage, and thus minimize the cost and performance impact.
Below are a few things we plan to add to the feature before the GA:
- Allow customers to enable/disable telemetry on a per API basis
- Add sampling controls to allow customers to specify sampling rate
- Allow customers to emit custom Traces telemetry items using a policy