Skip to content
Workloads — one page per Fabric workload with parity grade and honest gaps

CSA Loom — Workloads

For every Microsoft Fabric workload (and Fabric-only capability), one documentation page explains how CSA Loom delivers parity, where the honest gaps are, and how the workload migrates forward when Fabric reaches your audit boundary.

Storage + namespace

  • OneLake parity

    ADLS Gen2 + unified path tree + cross-cloud shortcuts service. Engine-layer enforcement (vs Fabric's storage-protocol).

Data Engineering + Warehouse + Data Science

  • Data Engineering

    Databricks lakehouse + Spark notebooks + environments + Materialized Lake Views (DLT in Commercial; scheduled Jobs in Gov).

  • Data Warehouse

    Databricks SQL Warehouse (Commercial) or Synapse Serverless (Gov). T-SQL DML over Delta; cross-warehouse queries.

  • Data Science

    Databricks notebooks + MLflow + Vector Search (Commercial) or Azure AI Search vector (Gov) + custom AI Functions library.

Real-Time + Activator

  • Real-Time Intelligence

    Azure Stream Analytics + ADX (same engine as Fabric Eventhouse) + KQL Querysets + ADX dashboards.

  • Data Activator parity (Reflex)

    Loom Activator Engine: NRules + Redis state + Function dispatcher backed by ADX. All 8 Fabric Reflex primitives.

Mirroring + Direct Lake

  • Mirroring parity

    OSS Debezium + Spark Structured Streaming + Delta MERGE. Honors Fabric's Open Mirroring publisher contract.

  • Direct Lake parity

    Power BI Premium Import + warm-cache materializer (5-30s freshness). Honest gap vs Fabric's sub-second documented openly.

Agents + Copilot

  • Data Agents parity

    Loom Data Agents extends apps/copilot/ with NL2SQL / NL2DAX / NL2KQL tools + per-source few-shot examples + OBO identity.

  • Copilot in CSA Loom

    Per-pane Copilot personas (notebook, warehouse, DAX, KQL, activator, agent-config). Built on Loom Data Agents runtime.

Future workloads

  • Fabric IQ family

    Ontology / Graph / Plan / Operations Agent / Maps — v2 deferred (Operations Agent ships in v1.1).

Parity matrix at a glance

See Parity matrix for the keystone summary table across all workloads.

How to read each workload page

Each page follows this structure:

  1. What Fabric does — one-paragraph summary of the Fabric capability
  2. CSA Loom parity design — architecture + implementation notes
  3. Per-boundary behavior — Commercial / GCC / GCC-High / IL5
  4. Honest gaps — what we can't match; why
  5. Forward migration — to Fabric when Gov GA arrives
  6. Related — PRP that builds it, tutorial that demonstrates it, runbook for incidents

This ensures you can read any workload page in isolation and understand the parity story end-to-end.