Skip to content
CSA Loom — the Microsoft Fabric experience for Azure tenants where Fabric isn't yet available: lakehouses, warehouses, notebooks, semantic models, Activator rules, Data Agents, across Commercial, GCC, GCC-High, and DoD IL5

Mirroring parity

Comparative positioning note

This document is written from the perspective of Microsoft Azure, Cloud Scale Analytics, and CSA Loom. Any description of third-party or competing products, services, pricing, or capabilities is derived from publicly available documentation and sources believed accurate at the time of writing, and is provided for general comparison only. We do not claim expertise in, or authority over, any non-Microsoft product or service; the respective vendor's official documentation is the authoritative source for their offerings, which may change over time. Nothing here is intended to disparage any vendor — where a competing product has genuine advantages, we aim to note them honestly. Verify all third-party details against the vendor's current official documentation before making decisions.

What Fabric does

Zero-ETL near-real-time CDC into OneLake as Delta tables. GA sources as of 2026-05-22: Azure SQL DB, Azure SQL MI, SQL Server 2016-2025, Cosmos DB, Azure DB for PostgreSQL, Snowflake, Oracle, SAP Datasphere, Fabric SQL DB.

Fabric publishes the Open Mirroring landing-zone protocol so partners can drop Parquet files with a documented __rowMarker__ column directly into a landing zone path — Fabric's replicator picks them up and MERGEs into Delta. This is the partner-extensible ingestion path (Qlik Replicate, Striim, Informatica IDMC, SNP Glue for SAP, Theobald Xtract Universal).

Latency: sub-minute steady-state; sub-second via partner streaming.

CSA Loom parity design — apps/fiab-mirroring-engine

Per ADR fiab-0006: OSS Debezium + Spark Structured Streaming + Delta MERGE. Honors Fabric's Open Mirroring publisher contract so partner publishers work unchanged.

Source connectors

Source Mechanism
Azure SQL DB / Azure SQL MI Debezium SQL Server connector (CDC tables)
Postgres Debezium Postgres connector (logical replication)
MySQL Debezium MySQL connector (binlog)
Cosmos DB Azure Cosmos Spark connector (change feed)
Snowflake Custom poller via Snowflake streams API
Oracle Debezium Oracle (LogMiner)
SQL Server 2016-2025 on-prem Debezium SQL Server + Self-Hosted IR
SAP / partner-published Parquet Open Mirroring landing-zone protocol

Transport

Event Hubs (Kafka protocol surface) — Debezium emits Kafka topics; Event Hubs accepts them natively. Cross-cloud Kafka egress not required.

Replicator

Spark Structured Streaming job on Databricks — reads Event Hubs + landing zone, parses CDC envelope (op = c/u/d/r, before, after), MERGEs into Delta target idempotently. Trigger interval 30 s default (configurable down to 5 s).

Open Mirroring landing-zone protocol

Identical to Fabric's:

  • Path: <ADLS>/landing-zone/<schema>/<table>/
  • _metadata.json declares keyColumns
  • 20-digit zero-padded sequence file names (00000000000000000001.parquet)
  • __rowMarker__ column: 1=INSERT, 2=UPDATE, 3=DELETE semantics

Per-mirror configuration

Stored in Cosmos DB. Authored via Loom Console "Mirroring" pane (v1.1) or JSON-direct in v1:

{
  "id": "mirror-finance-sales-azuresql",
  "workspaceId": "ws-001",
  "source": {
    "type": "azure-sql-db",
    "server": "sqlsrv-finance-prod.database.windows.net",
    "database": "SalesDB",
    "tables": ["dbo.customers", "dbo.orders"],
    "credentialKVRef": "https://kv-loom.vault.azure.us/secrets/finance-sql-cdc"
  },
  "target": {
    "lakehouseId": "finance-lakehouse",
    "tablePrefix": "raw_sales_",
    "deltaProperties": {"delta.enableChangeDataFeed": "true"}
  },
  "triggerIntervalSeconds": 30,
  "schemaEvolution": "auto-union",
  "enabled": true
}

Per-boundary behavior

Boundary Compute Event Hubs Kafka Spark Streaming
Commercial / GCC Container Apps ✅ Databricks
GCC-High / IL4 AKS ✅ Databricks classic
IL5 (v1.1) AKS ✅ Databricks classic

All Gov boundaries have Event Hubs + Databricks (classic clusters) authorized.

Honest gaps

  • First-touch setup UX harder than Fabric's "click to mirror" — customer configures Debezium connector + Spark job parameters; v1 ships templated configs per source type; v1.1 polishes UX
  • Snowflake source has no native Debezium — custom poller is more fragile; document operational expectations openly
  • Latency — sub-minute steady-state matches Fabric's typical; sub- second possible via partner streaming but harder than Fabric's managed-path

Forward migration

When Fabric Mirroring lands in Gov, customers can keep Loom Mirroring running OR switch per-source. Since both write Delta to the same logical lakehouse path (via OneLake shortcut), switching is per-source case-by-case.