FabOps Governance

Inspiration

I work with Microsoft Fabric since its debut. It's a great platform, but it needs better governance. That's what's inspired this work. I had already built many scripts for Fabric governance, but I always needed something more automated.

What it does

This system has a definition of a new language for governance. The agent helps you build new governance rules using this language.

For example: Each workspace needs to have two groups as admins. The lakehouses needs to be using schemas.

The system saves this rules using elastic. Elastic also saves the results of the rules execution.

After the rules are saved, the system can execute them.

How I built it

I build a systme of agents with orchestration surrounding them. The rules need to be created, managed in elastic, compiled to an execution format and executed, with the results saved in elastic as well. At the moment, this uses 3 agents for these tasks.

Challenges I ran into

I started building in AI Studio, but the AI Studio environment is very limited. After facing many limitations, I had to use the ADK to deploy the agents to Agent Platform

Accomplishments that I'm proud of

This is a starting point to a very solid solution and it was built very fast.

What I learned

It was one of the most complex orchestrations I built and it was my first contact with Google Cloud

What's next for FabOps Governance

The rules system needs to be improved and we need to include additional execution modes to allow more complex rules to be processed.

Built With

• adk
• ai
• elastic
• python