AI-200
Azure
Back to Observe and troubleshoot apps on Azure

Analyze app telemetry with logs and metrics

Create workbooks for interactive analysis

Overview

Azure Workbooks are interactive reports inside Azure Monitor — parameterized KQL, charts, grids, and markdown in one document. Use them when dashboards are not enough and you need guided analysis of AI app telemetry.

Exam tips

  • Workbooks support parameters (time range, operation name, service) that feed queries
  • Combine text + queries + visualizations for runbooks
  • Save to a resource group; share with RBAC
  • Great for “why is RAG slow?” style investigations across requests + dependencies

When to use

Use a workbookPrefer a plain dashboard / saved query
Multi-step investigation with filtersSingle KPI glance
Shared troubleshooting runbookAd-hoc one-off query
Comparing services via parametersPinning one chart

Workbook

An Azure Workbook is an interactive canvas of markdown, parameters, and query visualizations backed by Logs, Metrics, or Azure Resource Graph. Authors build analysis stories; operators change parameters without rewriting KQL.

Typical structure for an AI API workbook:

  1. Parameters — time range, service.name, HTTP route
  2. Overview — request volume, failure rate, P95
  3. Dependencies — slow or failing outbound calls
  4. Exceptions — top ProblemIds linked by OperationId
  5. Sample traces — grid of recent failed OperationIds to copy into Transaction search

Parameters and queries

Parameters are workbook inputs (dropdowns, text, time) referenced inside queries so one workbook serves many scenarios.

Example pattern (conceptual):

Kql
AppRequests
| where TimeGenerated > ago(24h)
| where Name has "{OperationName}"   // bound to workbook parameter
| summarize
    Total = count(),
    Failures = countif(Success == false),
    P95 = percentile(DurationMs, 95)
  by bin(TimeGenerated, 1h)

Add a second query step that uses selected OperationId from a grid to show the full union timeline.

Create a workbook (portal)

  1. Azure Monitor or Application Insights → Workbooks
  2. Empty / use a template
  3. Add parameters (TimeRange, text for route/model)
  4. Add query steps against your App Insights / workspace
  5. Choose visualization (time chart, grid, tiles)
  6. Save to a resource group with a clear name (AI-200 RAG health)

Tips for AI workloads

  • Parameterize model deployment name via customDimensions
  • Include a step for vector DB dependency targets (Cosmos / PostgreSQL)
  • Link markdown “what to check next” for on-call engineers
  • Keep queries scoped with ago(...) / TimeRange to control cost

Learn more