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

Loom Direct-Lake Shim service

Per ADR fiab-0004 and Direct Lake parity workload.

Purpose

Best-effort Direct Lake parity. Keep Power BI Premium semantic models refreshed close to the underlying Delta commit cadence using notification-driven partition-scoped refresh.

Latency target: 5-30 seconds median from Delta commit to refreshed model for partition-aware tables. Honest gap vs Fabric's sub-second documented openly.

Service shape

Aspect Value
Repo path apps/fiab-direct-lake-shim/
Language C# .NET 10
TOM client Microsoft.AnalysisServices.Core 19.x
Trigger Azure Event Grid on Storage BlobCreated events
State store Azure Cosmos DB (per-table version map)
Concurrency control Azure Cache for Redis distributed lock
Container host Container Apps (Commercial / GCC); AKS (GCC-H / IL5)
Build PRP PRP-08

Architecture

   Customer lakehouse                  Loom Direct-Lake Shim
   ───────────────────                 ───────────────────────────
   ADLS Gen2 Delta                     Container App / AKS workload
   _delta_log/*.json
        │
        │ Event Grid: BlobCreated
        │ filtered to /_delta_log/*
        ▼
                                    ┌──────────────────────────────┐
                                    │ EventGridTrigger handler     │
                                    │ 1. Parse Delta commit info   │
                                    │ 2. Identify changed parts    │
                                    │ 3. Resolve to semantic-model │
                                    │    partitions                │
                                    │ 4. Acquire Redis lock        │
                                    │ 5. TOM partition refresh     │
                                    │    via XMLA endpoint         │
                                    │ 6. Update version tracker    │
                                    └──────────────────────────────┘
                                              │
                                              ▼ refresh via XMLA
                          Power BI Premium semantic model
                          (Premium F-SKU capacity)

Per-table refresh policy

Stored in Cosmos DB; editable via Console "Semantic Model" pane (v1.1) or CLI (v1):

Policy Behavior
partition Refresh only the affected partition
full Full table refresh
directquery-fallback DirectQuery against Synapse Serverless / Databricks SQL
composite TMDL composite (mix of Import + DirectQuery)

Health endpoint

{
  "status": "healthy",
  "event_grid_subscription": "active",
  "tracked_tables": 23,
  "last_refresh_age_seconds": 18,
  "median_refresh_latency_seconds": 14,
  "p95_refresh_latency_seconds": 27
}

Operational SLAs

Metric Target
Refresh latency p50 (partition) < 30 s
Refresh latency p95 (partition) < 60 s
Refresh latency p50 (full table, small) < 90 s
Concurrent refresh deadlock prevention Redis lock — no double-refresh
Event Grid → handler latency < 5 s

Limitations

  • Sub-second freshness not achievable — see Direct Lake parity
  • Power BI Premium F-SKU required — Pro / Premium-Per-User not supported (XMLA endpoint requires Premium)
  • GCC has no F-SKU — Direct Lake parity unavailable in GCC (structural gap)

Runbooks