Azure DevOps offers a repository health feature, which allows to monitor multiple metrics that contribute to the health of your Git repositories. If you are using Azure DevOps services, you maybe already know this feature. But with the latest release of Azure DevOps server (December 2025) , it finally arrived on-premise as well. Reasons enough to have a deeper look at it and write this post.
Let’s dive in!
Why is this feature relevant for my team?
Think of your Git repository like a living organism. As it grows with more commits, blobs, branches, and objects, it can become sluggish and unwieldy. Large repositories increase the load on Azure DevOps infrastructure, affecting performance and user experience. Without proper maintenance, your team could face slower clone times, degraded git operations, and even service disruptions.
The repository health feature now provides visibility into key health metrics and offers actionable recommendations before problems escalate.
Getting to your repository's health dashboard is straightforward. Navigate to your Azure Repos Git repository, select Repo > Files, and choose "Health and usage" from the ellipsis menu.
This opens the Repository health and usage panel, where you'll find a comprehensive overview of your repository's vital statistics.
The interface uses a traffic light system: factors highlighted in red indicate unhealthy conditions that need immediate attention, while amber warnings suggest you're approaching problematic thresholds.
Key metrics
The health feature tracks the following key metrics:
Repository size
Microsoft recommends keeping your repository size under 100 GB for optimal performance. Smaller repositories clone faster and are easier to manage. If you're exceeding this threshold, consider using Git Large File Storage (Git-LFS), or Scalar to manage large repositories.
Binary and large files
Storing large binary files like images, videos, or compiled artifacts directly in Git is problematic. Different versions of these files don't compress efficiently through Git's delta compression, meaning each version consumes substantial disk space. The health dashboard helps you identify when these files are becoming a problem.
Object counts
Azure Repos has a hard limit of 100 million objects in a single repository. The health panel tracks several object types:
- Reachable blobs: The total size of file content stored in your repository. These should stay well under 100 GB, leaving room for other object types.
- Reachable trees: Tree objects represent directory structures. They grow as you add directories and files, and Git creates copies of trees whenever files change. Distributing files across multiple subdirectories rather than having flat directory structures helps keep tree counts manageable.
- Reachable commits: The total number of commit objects in your repository. While commits are smaller than blobs, a large number can slow down history traversal operations.
Reference count
If your Git repository contains more than 10,000 refs, you should consider enabling Limited Refs. As reference counts increase, so does the data that needs to be negotiated between client and server, potentially degrading user experience.
Best practices for repository health
Based on the health metrics Azure DevOps monitors, here are some best practices to keep your repositories in top shape:
- Regularly review your health dashboard: Make repository health checks part of your routine maintenance, especially for high-activity repositories.
- Refactor large files: Move binary assets, build artifacts, and other large files to appropriate storage solutions like Azure Artifacts or Git-LFS.
- Maintain reasonable directory structures: Avoid deep nesting or overly flat hierarchies. A balanced approach reduces tree object proliferation.
- Archive stale branches: Old feature branches and experimental work can be archived or deleted to reduce ref counts.
- Consider repository splitting: If a repository has grown beyond manageable size, evaluate whether it makes sense to split it into multiple repositories organized by logical boundaries.
More information
Repository health - Azure Repos | Microsoft Learn
Azure DevOps Server Release Notes - Azure DevOps Server & TFS | Microsoft Learn
Limited Refs - Azure DevOps Server & TFS | Microsoft Learn
Work with large files in your Git repo - Azure Repos | Microsoft Learn