Introduction to Workflow Tracking in .NET Framework 4.0 Beta1

By now you must be aware of the significantly enhanced Windows Workflow Foundation (WF) scheduled to be released with .Net Framework 4.0. The road to WF 4.0 and .Net Framework 4.0 Beta1 documentation for WF can give you more details. Being a member of the team responsible for the development of the WF tracking feature, I am excited to discuss the components that constitute this feature. In a nutshell, tracking is a feature to gain visibility into the execution of a workflow. The WF tracking infrastructure instruments a workflow to emit records reflecting key events during the execution. For example, when a workflow instance starts or completes tracking records are emitted. Tracking can also extract business relevant data associated with the workflow variables. For example, if the workflow represents an order processing system the order id can be extracted along with the tracking record. In general, enabling WF tracking facilitates diagnostics or business analytics over a workflow execution. For people familiar with WF tracking in .Net 3.0 the tracking components are equivalent to the tracking service in WF 3. In WF 4.0 we have improved the performance and simplified the programming model for WF tracking feature.

A high level view of the tracking infrastructure is shown below

The primary components of the tracking infrastructure are

1) Tracking records emitted from the Workflow runtime.

2) Tracking Profile to filter tracking records emitted from a workflow instance.

3) Tracking Participants that subscribe for tracking records. The tracking participants contain the logic to process the payload from the tracking records (e.g. they could choose to write to a file).

The Workflow tracking infrastructure follows the observer pattern. The workflow instance is the publisher of tracking records and subscribers of the tracking records are registered as extensions to the workflow. These extensions that subscribe to tracking records are called tracking participants. Tracking participants are extensibility points that allow a workflow developer to consume tracking records and process them. The tracking infrastructure allows the application of a filter on the outgoing tracking records such that a participant can subscribe to a subset of the records. The mechanism to apply a filter is through a tracking profile.

The workflow runtime is instrumented to emit tracking records to follow the execution of a workflow instance. The types of tracking records emitted are

· Workflow instance tracking records: Workflow instance records describe the life cycle of the workflow instance. For instance a record is emitted when the workflow starts or completes.

· Activity tracking records: Activity tracking records are emitted when a workflow activity executes. These records indicate the state of a workflow activity (i.e. an activity is scheduled, activity completes or fault is thrown).

· Bookmark resumption tracking record: A bookmark resumption record tracks any bookmark that is successfully resumed

· User tracking records. A workflow author can create custom tracking records within a custom workflow activity and emit them within the custom activity. Custom tracking records can be populated with data to be emitted along with the records.

Out of the box in WF 4.0 we provide an ETW (Event Tracing for Windows) based tracking participant. The ETW tracking participant writes the tracking records to an ETW session. The participant is configured on a workflow service by adding a tracking specific behavior in a config file. Enabling ETW tracking participant allows tracking records to be viewed in the event viewer. Details of using the ETW based tracking participant will be covered in a future post. The SDK sample for ETW based tracking is a good way to get familiar with WF tracking using the ETW based tracking participant.

In future posts we will discuss the WF tracking feature in depth. Topics covered will include tracking profiles and tracking records, ETW tracking participant, writing custom tracking participants, variable extractions and unified tracking and tracing.