Updates for the Innovate and Modernize Apps with Data and AI MCW

A little while ago, I talked about updating the Innovate and modernize apps with Data and AI Microsoft Cloud Workshop. Here’s a quick update with an overview of the changes.

A Simpler Architecture

Here was the architecture diagram prior to the update:

This all makes perfect sense, trust me.

Here’s the new diagram:

This therefore makes perfecter sense?

You can see that the diagram itself is considerably simpler, and there are a few things which helped here:

  • Azure Database for PostgreSQL is gone from the lab. There’s a place for a lab with Postgres, but this wasn’t a great example of it.
  • “Informational” links have been removed, including Cosmos OLTP to OLAP.
  • Fewer moving parts. The old diagram had data movement going in all directions. In this case, the data flow is almost always going from left to right, with the exception being the microservice for metadata at the end.

…Leading to Fewer Exercises

By removing Postgres from the equation, I was able to eliminate one exercise entirely and strip down another one, saving a good amount of time.

The other big change was one which saddened me a bit: removing IoT Hub from the equation. I was really proud of the process for setting up IoT Hub and IoT Edge, assigning a device to Edge, and emulating an IoT device. In the lab itself, the factory and IoT Hub components are now “integrated” into the Azure Function, which generates artificial telemetry data and writes it to Cosmos DB. I thought the IoT part was an exciting component of the lab, but at the same time, it takes about 1 hour to get it going and the Powers That Be preferred that we get to the heart of the exercise (Cosmos, Synapse Link, and microservice deployment) sooner.

More Troubleshooting

One area people have had problems in the past is in creating and deploying the Azure Machine Learning model. For this reason, I’ve added some troubleshooting steps to give learners a chance to resolve issues. You can get most of the relevant information from the Azure Machine Learning logs, assuming the container gets built and logging catches the issue. But those are not always reasonable assumptions.

Conclusion

Even stripping out the sections on IoT Hub, this is still one of the most comprehensive end-to-end MCWs out there. This is free content that trainers can develop full-day trainings around, with a couple hours for whiteboard design and 3-4 hours in the lab (previously, it was about 6 hours if nothing went wrong). So give this MCW a try and if you have any suggestions or run into problems, please do add a GitHub issue.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s