Reference architecture diagram

  • Article

Appropriate roles: Co-sell solution admin

A reference architecture diagram is a model of the infrastructure that your Microsoft commercial marketplace solution relies on.

For Azure intellectual property (IP) solutions, the diagram should also show how your solution uses Microsoft's cloud services per the technical requirements of IP co-sell.

A reference diagram isn't intended to assess the quality of an architecture. It's intended to show how your solution uses Microsoft services.

You can create a reference architecture diagram using various tools. However, we recommend Microsoft Visio because it has Azure Architecture models that you can use as a helpful starting point.

Typical components of a reference architecture diagram

Your reference architecture diagram must clearly identify your IP as a solution, application, or service code that is both deployed on and driving consumption of Microsoft Azure.

Your code must be highly reusable and not depend on extensive customization per deployment.

The diagram must provide a good understanding of how your solution uses Microsoft services. The diagram must include the following:

  • A high-level description of the data flow providing an understanding of the interactions between components of the diagram
  • Logical boundaries for Azure cloud, customer cloud, on-premises environment, and integration with other cloud services
  • Azure cloud services that host and interact with your solution, including ones that consume Azure resources
  • Identification of repeatable Intellectual Property (IP) code; the code you are providing as part of the solution
  • User interfaces and other services exposing the solution
Example reference architecture diagram: vertical industry chatbot

This image is an example Co-sell architecture diagram.

The image is an example of a reference architecture diagram. It's a diagram of a vertical industry chatbot that can be integrated with intranet sites to help forecast demand scenarios using a machine learning algorithm. It uses supply-chain and manufacturing-schedule data from various enterprise resource planning (ERP) systems. The bot is designed to address questions about when a salesperson can commit to possible delivery dates for an order.

Next steps