Warning: Fabric notebooks may share compute sessions across pipelines
In Microsoft Fabric, notebooks run from different pipelines can end up sharing the same computing session. This means data, variables, or state from one pipeline run could leak into another. It's a subtle gotcha worth knowing before it causes hard-to-debug problems.
Microsoft Fabric is a cloud platform for running data workflows. When you set up multiple pipelines — automated sequences of tasks — that each trigger the same notebook, you might assume each run gets its own isolated environment. That assumption can be wrong.
Fabric supports a feature called High Concurrency (HC) sessions, which allows multiple pipelines to share a single running session to save resources. The side effect is that variables or state set during one pipeline's execution may still be present when another pipeline runs in the same session. This can cause unexpected errors or incorrect results that are difficult to trace. If your pipelines need to run independently, check your session isolation settings to make sure HC session sharing is disabled or properly scoped.
Key points
- Fabric notebooks can share a High Concurrency (HC) session across multiple pipelines
- Shared sessions mean variables or state from one run can affect another
- This can cause subtle, hard-to-find bugs in automated workflows
- Check session isolation settings if pipelines must run independently
- Review Fabric documentation on HC session behavior before deploying parallel pipelines
Quick term guide
- Microsoft Fabric
- A Microsoft platform that combines data management and AI integration tools in one place.
- pipeline
- An automated sequence of steps that processes or moves data without manual intervention.
- workflows
- The specific order of steps taken to finish a piece of work.
- workflow
- A repeatable set of steps for getting a task done.
- trigger
- A signal or condition that starts a task.
- High Concurrency (HC) session
- A shared computing environment that lets multiple tasks run at the same time using the same resources.
- sources
- Evidence showing where a piece of information came from.
- session isolation
- A setting that ensures each pipeline run gets its own separate environment, preventing data from leaking between runs.