Microsoft Fabric Architecture
Fabric puts workloads on one lake and, by default, one capacity. See what that buys, where the pool becomes a failure domain, and what a Synapse move costs.
-
-
Throttling
Throttling slows or refuses requests once a consumer passes its limit, so one burst does not take down everyone. Server-side limits protect capacity by themselves; callers that are told and back off keep retries from adding to the load.
-
Mirroring (Microsoft Fabric)
Mirroring replicates an operational database into OneLake continuously, with no pipeline to build. It is still a copy, with lag and retention to manage.
-
Direct Lake
Direct Lake reads Delta tables in OneLake instead of importing them, so a refresh updates only metadata. Guardrails, fallback, and security decide the rest.
-
OneLake
Every Fabric tenant has exactly one OneLake. Workloads store and read Delta or Iceberg tables in it, so one copy of data can serve Spark, SQL, and Power BI.
-
OneLake Shortcut
A shortcut makes an S3 bucket or another workspace’s table look like a local folder in Fabric. No separate copy is maintained, though optional caching can store files locally, and deletes and access can reach the source.
-
Capacity Unit (Microsoft Fabric)
An F64 capacity provides 64 capacity units, shared by every workspace assigned to it. The number sets compute, licensing rules, and who slows down whom.
-
Capacity Bursting and Smoothing (Microsoft Fabric)
A spike can run faster than the capacity you bought, because Fabric spreads the extra usage over the next minutes or hours. The borrowed compute still counts.