Practical write-ups on Microsoft Fabric, Azure Data Factory, data engineering, Power BI, cloud architecture and data warehousing. No gated PDFs.
On a migration with several hundred source tables, hand-writing a notebook per table is the obvious approach and the wrong one. Here is the configuration-driven pattern we use instead, and the audit trail you get for free.
When a bill jumps, the instinct is to resize the cluster. In our experience the cause is usually thousands of tiny files per partition. Check file counts before touching cluster configuration.
The parallel run is always first to be cut under schedule pressure. Its purpose is not to prove the pipelines work — tests do that. It is to surface the reports nobody told you about.
One pipeline, one configuration table, any number of sources. How to structure the control table, handle incremental watermarks, and keep reruns idempotent.
Most Power BI performance problems are modelling problems. Star schemas, the cost of bi-directional relationships, and when Direct Lake is genuinely the right answer.
How to reconcile a migrated warehouse against its predecessor — counts, balances, and the edge cases that only appear at period end.
Occasional, practical, and never more than once a month. Unsubscribe any time.
Tell us what you are running today and where it hurts. You will get an honest view of the options, with no obligation and no sales script.