KQL: where, project, summarize, bin(), join and render against requests, dependencies, exceptions and traces

Kusto Query Language (KQL) is how you query Application Insights and Log Analytics data. A query starts with a table name and pipes rows through operators, each separated by |, much like a Unix shell pipeline. The tables you will use most from Application Insights are requests (incoming calls to your app, with name, url, resultCode, success and duration), dependencies (outgoing calls, with target, type, success and duration), exceptions (errors, with type, outerMessage and problemId) and traces (log messages, with message and severityLevel). All share timestamp, operation_Id and cloud_RoleName. In a workspace-based resource queried directly from Log Analytics, the same data appears in tables named AppRequests, AppDependencies, AppExceptions and AppTraces with slightly different column names.