Building a real-time demo in Microsoft Fabric takes an afternoon. Running it in production is another story. In the demo, the source never goes offline and every event arrives exactly on time.

In this session I start with a quick tour of what Fabric offers for real-time data and what a typical setup looks like: events flow in through an Eventstream, land in an Eventhouse where you can query them within seconds, and end up on a real-time dashboard or trigger an alert. Then I share what we learned taking pipelines like that into production: deploying real-time items across environments, monitoring that tells you when data stops arriving, recovering from source outages, and what an always-on workload does to a shared capacity.

Then we break things live. We start with a pipeline that works, switch off the source, let the data go stale and send events late. After each failure we add the piece that would have caught it, until the pipeline alerts you, recovers by itself and fills the gap without duplicates.

You will leave this session with:
- A way to monitor data freshness on event time, so you hear about missing or late data before your users do
- A pattern for restarting and backfilling after an outage without losing or duplicating records
- A setup for deploying real-time solutions across development, test and production
- A go-live checklist you can use on your own project the next morning