Analyze pipeline load to determine agent configuration and capacity

In this lesson, you will learn how to analyze Azure Pipelines load and optimize agent configuration and capacity. Whether you’re preparing for the AZ-400 certification or streamlining your CI/CD workflows, understanding agent pools, concurrency, and performance metrics is key to efficient, cost-effective builds and deployments.

1. Agent Types and Pools in Azure Pipelines

Azure Pipelines agents execute your jobs. There are two primary types:

Microsoft-hosted agents: Preconfigured and maintained by Microsoft, but limited in customization.
Self-hosted agents: Fully under your control—OS updates, scaling, and security are your responsibility.

Agents are grouped into agent pools, which you share across projects and pipelines.

Note

Self-hosted agents offer greater flexibility but require you to handle maintenance, OS patches, and infrastructure scaling.

Viewing Agent Pools

In Azure DevOps, navigate to Project Settings > Agent pools.
Review each pool’s agents and their status (online, idle, offline).

The image shows the Azure DevOps interface, specifically the "Agent pools" section under "Project Settings," displaying details of a hosted agent that is currently online and idle.

2. Analyzing Microsoft-hosted Agent Usage

To inspect Microsoft-hosted agent performance:

Go to Pipelines > Jobs.
Examine queue time, wait time, and duration for each job.

The image shows a screenshot of the Azure Pipelines interface, displaying a list of jobs with their status, project names, agent specifications, queue times, wait times, and durations.

3. Monitoring Pool Consumption and Concurrency

The Pool consumption dashboard provides insight into concurrent job usage, queued jobs, and running jobs. More concurrency reduces queue time but increases cost.

The image shows a dashboard from Azure Pipelines displaying pool consumption reports for public and private hosted concurrency, with graphs indicating concurrency, queued jobs, and running jobs over time.

4. Using Self-hosted Agents

Under Agent pools > Default, register your own Windows, Linux, or container-based agents. These can run on-premises or in any cloud subscription, giving you full control over VM size and software dependencies.

5. Key Pipeline Performance Metrics

Track these metrics to right-size your agents and pipelines:

Metric	Definition	Benefit
Queue time	Time a job waits before an agent picks it up	Spot peak-load bottlenecks
Job execution time	Duration from job start to completion	Identify slow build or test steps
Parallel builds	Number of jobs running simultaneously	Optimize agent utilization

6. Viewing Pipeline Analytics

Navigate to Pipelines > [Your Pipeline] > Analytics for:

Pipeline Pass Rate
Test Pass Rate
Pipeline Duration

Click Pipeline Duration to view overall run times and the top 10 slowest steps.

The image shows an Azure DevOps pipeline duration report, displaying a graph of pipeline duration over time and a breakdown of the top 10 steps by duration.

In this example, checkout and initialize job are the longest steps. Optimizing these can significantly reduce total run time.

7. Comparing Pipeline Step Durations

Compare multiple pipelines to spot heavy tasks like VSTest or VSBuild.

The image shows a pipeline duration report from Azure DevOps, displaying a graph of pipeline durations over time and a breakdown of the top 10 steps by duration.

Long-running test or build phases often benefit from more CPU, memory, or additional self-hosted agents.

8. Managing Parallel Jobs and Capacity

Under Project Settings > Parallel jobs, view your subscription’s limits:

Tier	Parallel jobs	Free minutes/month	Notes
Free	1	1,800	Suited for individual developers
Paid	Up to 10+	Unlimited	Scale with purchased concurrency
Self-hosted	Unlimited	N/A	Only limited by your hardware

Warning

Hitting your parallel job or minute quota will pause pipeline execution. Plan upgrades or add self-hosted agents proactively.

9. Custom Dashboards and Widgets

Aggregate metrics on a single pane:

Go to Overview > Dashboards.
Click Add widget, search for “Pipeline.”
Add Build History, Release Pipeline Overview, or community widgets from the Extension Marketplace.

The image shows an Azure DevOps dashboard with pipeline monitoring widgets, displaying build history and performance metrics. The interface includes options to add widgets and view pipeline statistics.

Monitoring Average Duration trends helps determine when to scale agents.

10. Integrating with Azure Log Analytics

For advanced telemetry, stream pipeline logs to an Azure Log Analytics workspace. Query build and release metrics over time for root-cause analysis.

The image shows a Microsoft Azure portal page for creating a Log Analytics workspace, with fields for project and instance details.

Learn more in the Azure Monitor Logs documentation.

11. Provisioning Additional Build Agents

To scale quickly, install the Build Agents for Azure DevOps solution from the Azure Marketplace. It automates VM provisioning and agent registration:

Select your subscription, resource group, and region.
Choose VM size (vCPUs, RAM).
Deploy and connect to your agent pool.

The image shows a Microsoft Azure portal interface for creating a single Windows VM plan, with fields for subscription, resource group, region, virtual machine name, and credentials. A warning message indicates that the selected resource group contains existing resources.

The image shows a Microsoft Azure portal interface for selecting a virtual machine (VM) size, displaying various options with details like vCPUs, RAM, and data disks. The current selection is a Standard B1ms VM with 1 vCPU and 2 GB memory.

Alternatively, create VMs manually or via your own ARM/Bicep templates.

Conclusion

Tracking queue time, execution time, and concurrency enables you to right-size Azure Pipelines agents and avoid bottlenecks. Use built-in analytics, custom dashboards, Log Analytics, and Marketplace extensions to monitor performance and scale your CI/CD infrastructure on demand.

Links and References

Watch Video

Watch video content