Module 6 of 6 · ~3 minutes

Tracking

What Tracking is

Tracking is a historical, searchable view of every step of every integration that has run on the platform. When a run has finished, Tracking is what lets you open it back up and see exactly what happened, one step at a time.

Keep two things apart. The runtime always keeps a thin internal record of in-flight messages, so a suspended run can be resumed, a running run can be observed, and an errored run shows up where you expect. That lightweight record is always on. Tracking is the historical layer on top of it: opt-in per port, and once enabled it persists the full step-by-step record of each run for later search and analysis.

Levels: set per port

Tracking is turned on per port, not per integration, and each port carries one tracking-level setting with three values, from lightest to most detailed:

Only on Error. Nothing is written for successful runs. On failure, the full step history is persisted, including the message body at each step. The cheapest level when failures are rare.
Enabled. Every run through the port is recorded (timestamps, step identifiers, message type, adapter). Message bodies are not stored.
Enabled + Body. Every run is recorded including the full message body at each step. The most diagnostic, and the most expensive in storage.

Set the level on every port a flow touches. A flow whose three ports sit at three different levels produces a record that is detailed where one port enabled it and silent where the others did not. To see a run end to end, set the level uniformly across every port in the flow.

The Tracking screen

One screen, three regions. Five status boxes at the top, each with a live count: Successful, Running, Suspended, Exceptions, Errors. A filter strip (Tracking ID, start and end date, message type). And a results grid, one row per run, expandable to the full step-by-step history.

The boxes are also where you act. The actions match on a single row or on the whole count at once: Running can be Suspended or Terminated; Suspended can be Resumed or Terminated. Successful, Exceptions, and Errors are read-only. A bulk action against Running is best-effort, since runs can finish or move on between the click and the runtime handling it.

One run, one Tracking ID

Every run is stamped with a single Tracking ID the moment it enters the platform, and that same ID travels through every port the run touches: receive port, bus, send ports, loopback, null. Each port records its own steps under that shared ID, and the Tracking screen uses it to reassemble the steps into one logical, end-to-end run. Filter on the Tracking ID and you can recall the complete history of one specific run, even months later.

This is the part the platform calls out over other middleware, BizTalk included: not just that runs are tracked, but that every step is stitched into a single walkable run through one screen.

That's the course

Publish-select-deliver, the building blocks, the adapters, the habits, an AI assistant that keeps your data at home, and a tracking layer that lets you replay what ran. That is what the exam covers. The documentation goes far deeper when you are ready, with step-by-step tutorials from EDI orders to healthcare feeds.

Better yet, try it for real. The Starter edition is license free in the Azure Marketplace: installed, provisioned, and registered automatically in no more than 30 minutes, and you can start using it the moment you can log in.

Check yourself

Two quick questions on this module, in the same style as the exam. Not graded, not recorded, just practice with instant explanations.

Go deeper

Take the exam

The exam is free during the launch window. Once the window closes, a fee applies; certificates earned now stay valid.