Improving pre-aggregated Standard Metrics in SDK – Part II

[rajkumar-rangaraj]

Background

Part I - Improving pre-aggregated Standard Metrics in SDK

Default SDK configuration inspects every Request, Dependency telemetry items and aggregates metric dimension through AutocollectedMetricsExtractor telemetry processor before sampling processors. These pre-aggregated metrics are sent to Application Insights once every minute and are available to query under Application Insights standard metrics -> Server in metric explorer. As this metric is calculated before sampling, these metrics are superior in terms of accuracy.

Aggregates for Exceptions/Traces

Now, Exceptions and Traces standard metric dimension are calculated from ingested application insights data. If sampling is enabled metric dimension is calculated on sampled data, which is not accurate.

Plan is to provide accurate metric dimension CloudRoleName, CloudRoleInstance for Exceptions/Traces and real/synthetic, SeverityLevel for Traces. Users will be allowed to pick a max limit to the number of values. (similar to MaxDependencyTypesToDiscover)

AutocollectedMetricsExtractor telemetry processor to inspect Exception and Trace telemetry items and aggregate metric with key dimension before sampling processors. These pre-aggregated metrics are available to query under Application Insights standard metrics -> Server exceptions and Application Insights standard metrics ->Traces in metric explorer.

Server exceptions - Count of uncaught exceptions thrown in the server application.

Traces - Duration of trace calls.

[rajkumar-rangaraj]

@cijothomas