Hi @dbWizard
Thanks for bringing this to the Community!
Your bronze-lakehouse copy pattern is a solid way to insulate downstream analytics from a catastrophic mirrored-DB failure—but here are a few considerations and alternative ideas to weigh:
-
Making a 1:1 copy does safeguard reports and transforms, but it does incur extra storage and lineage overhead.
-
Incremental sync instead of full copy: Use Fabric Data Pipelines or Dataflow Gen2 with watermark parameters (RangeStart/RangeEnd) to only load changed or new rows, reducing cost and latency.
-
Snapshot isolation on the mirrored DB: If your mirrored database supports database snapshots or copy-only backups, you can snapshot to a durable store (OneLake, Blob) on a schedule, then point Bronze at those snapshots rather than re-copying live data.
-
Auto-failover groups or geo-replication: For critical mirrored DBs, consider Azure SQL’s auto-failover group feature to provide transparent DR and eliminate manual rebuilds.
-
OneLake shortcuts (if available): Instead of physically copying, use OneLake’s “shortcut” feature so Bronze points to a durable copy location outside the live mirror, yet doesn’t duplicate storage.
-
Monitoring & alerting: Whatever pattern you choose, build health checks (pipeline success, data freshness) to automatically detect and remediate sync failures before they affect consumers.
Each of these can reduce cost, simplify lineage, or provide more seamless DR—while still keeping Bronze as your single source for all downstream dependencies.
If this helps, please give us Kudos and mark the response as Accepted as solution.
Best Regards,
Community Support Team _ C Srikanth.
