πŸ‘€ Who is this for?

Solutions Engineer Data Architect IT Admin β€” Use these industry templates as starting points for Fabric architecture design. Each scenario includes recommended components, data flow, governance model, and SKU sizing.

Templates

Customer Scenario Templates

Pre-built architecture blueprints for common industry use cases. Click any scenario to expand the full blueprint.

πŸ“Œ These are starting points

These scenario templates are illustrative blueprints. Specific SKU sizing, component selection, and architecture patterns should be validated against your organization's requirements and current Fabric capacity guidance.

πŸ›’

Retail Analytics Platform

Retail F64+
β–Ύ

Unified analytics platform for multi-channel retail operations β€” combining POS transactions, e-commerce data, inventory, and customer behavior into a single Fabric lakehouse for real-time and historical insights.

πŸ—οΈ Architecture Components

Data Factory Lakehouse Data Warehouse Power BI Real-Time Intelligence Activator

πŸ”„ Data Flow

Sources
POS systems, e-commerce, inventory DB, loyalty program, supplier feeds
β†’
Bronze
Raw transactions, clickstream events, inventory snapshots (Delta)
β†’
Silver
Cleaned orders, deduplicated customers, joined inventory + sales
β†’
Gold
Sales metrics, customer 360, demand forecasting, store performance
β†’
Consume
Power BI dashboards, real-time inventory alerts, demand predictions

πŸ“ Recommended Sizing

Starting SKU F64
Why F64 Free Power BI viewers for store managers + Spark for daily ETL
Scale trigger Move to F128 when >200 concurrent BI viewers or >50 daily Spark jobs

πŸ”’ Governance Model

  • RLS per region/store: Store managers see only their location's data
  • Sensitivity labels: Customer PII (loyalty data) classified as Confidential
  • Workspaces: [Retail]-[Analytics]-[Dev/Test/Prod] naming convention
  • Domain: "Retail Operations" with merchandising and supply chain sub-domains
πŸ₯

Healthcare Data Platform

Healthcare F64+
β–Ύ

HIPAA-compliant data mesh architecture for healthcare organizations β€” separate domains for patient data, clinical operations, and financial analytics with strict access controls and audit trails.

πŸ—οΈ Architecture Components

Data Factory Lakehouse Data Warehouse Power BI Microsoft Purview Managed Private Endpoints

πŸ”„ Data Flow

Sources
EHR (Epic/Cerner), claims systems, lab results, pharmacy, wearables
β†’
Domains
Patient β€’ Clinical β€’ Financial (each with own lakehouse)
β†’
Medallion
Bronze→Silver→Gold per domain, PHI de-identified at Silver
β†’
Cross-Domain
OneLake shortcuts for population health and quality metrics
β†’
Consume
Clinical dashboards, regulatory reports, research datasets

πŸ“ Recommended Sizing

Starting SKU F64
Why F64 Managed VNets for HIPAA + enough CU for multi-domain ETL
Scale trigger Move to F128 for real-time patient monitoring or ML workloads

πŸ”’ Governance Model

  • HIPAA compliance: Managed private endpoints, no public internet access to data
  • PHI protection: Sensitivity labels auto-applied, CLS on SSN/MRN, dynamic masking
  • Audit trail: Purview lineage from source EHR to report, full access logs
  • Domain isolation: Patient, Clinical, Financial domains with separate security groups
🏦

Financial Services Reporting

Finance F128+
β–Ύ

Enterprise reporting platform for banks and financial institutions β€” regulatory compliance reports, risk analytics, and real-time trading dashboards with strong audit controls and SOX compliance.

πŸ—οΈ Architecture Components

Data Warehouse Lakehouse Power BI Deployment Pipelines Fabric Mirroring Microsoft Purview

πŸ”„ Data Flow

Sources
Core banking, trading platforms, risk engines, market data feeds
β†’
Ingest
Mirroring for real-time CDC from SQL databases + batch market data
β†’
Transform
Bronze→Silver reconciliation, Silver→Gold regulatory aggregations
β†’
Serve
Warehouse for regulatory SQL queries + Power BI dashboards

πŸ“ Recommended Sizing

Starting SKU F128
Why F128 Heavy SQL workloads for regulatory queries + concurrent analyst access
Scale trigger F256 for intraday risk calculations or >500 report consumers

πŸ”’ Governance Model

  • SOX compliance: Full deployment pipelines (Devβ†’Testβ†’Prod) with approval gates
  • CLS on sensitive fields: Account numbers and transaction amounts masked for non-privileged users
  • Private endpoints: No public access, VNet-integrated for regulatory requirements
  • Immutable audit: Purview lineage + activity logs retained per regulatory timelines
🏭

IoT & Manufacturing Analytics

Manufacturing F32–F64
β–Ύ

Real-time operational intelligence for manufacturing and IoT β€” streaming sensor data into Eventhouse for live monitoring, anomaly detection, and predictive maintenance, with historical analysis in Lakehouse.

πŸ—οΈ Architecture Components

Real-Time Intelligence Eventhouse Activator Lakehouse Spark Notebooks Power BI

πŸ”„ Data Flow

Sources
IoT sensors, PLCs, SCADA, MES, ERP (SAP/Oracle)
β†’
Stream
Event Hub / IoT Hub β†’ Eventhouse for real-time KQL queries
β†’
Process
Real-time anomaly detection + Activator alerts on thresholds
β†’
Store
Lakehouse for historical time-series, Spark for predictive ML
β†’
Consume
Live OEE dashboards, maintenance predictions, production reports

πŸ“ Recommended Sizing

Starting SKU F32
Why F32 Eventhouse ingestion is efficient; F32 handles moderate sensor volumes
Scale trigger F64 when ingesting >100K events/sec or adding ML training workloads

πŸ”’ Governance Model

  • Device-level access: Separate workspaces per plant/facility with dedicated security groups
  • Retention policies: Raw sensor data 90 days in Eventhouse, summarized in Lakehouse long-term
  • OT/IT separation: Network isolation between OT sensor networks and Fabric analytics
  • Alert governance: Activator rules reviewed quarterly, escalation paths documented
πŸ“ˆ

Marketing Analytics β€” Customer 360

Marketing F32–F64
β–Ύ

Unified customer view combining CRM, web analytics, social media, and campaign data β€” enabling identity resolution, attribution modeling, and personalized marketing with GDPR-compliant data handling.

πŸ—οΈ Architecture Components

Data Factory Lakehouse Spark Notebooks Data Warehouse Power BI OneLake Shortcuts

πŸ”„ Data Flow

Sources
CRM (Dynamics/Salesforce), GA4, social APIs, email, ad networks
β†’
Bronze
Raw events from each channel, API responses, campaign metadata
β†’
Silver
Identity resolution (Spark), unified customer profile, consent tracking
β†’
Gold
Customer segments, campaign attribution, lifetime value, churn scores
β†’
Consume
Campaign dashboards, audience exports, personalization APIs

πŸ“ Recommended Sizing

Starting SKU F32
Why F32 Identity resolution runs as batch Spark job; BI load is moderate
Scale trigger F64 when adding real-time personalization or >100M events/month

πŸ”’ Governance Model

  • GDPR compliance: Consent-based processing, right-to-deletion via Spark notebooks
  • PII masking: CLS hides email/phone from analysts, only aggregated segments visible
  • Data retention: Raw web events deleted after 13 months per GDPR, aggregates retained longer
  • Cross-team sharing: Marketing Gold layer shared via shortcuts with Sales domain (read-only)
🏒

Enterprise Data Platform

Enterprise F128+
β–Ύ

Full-scale enterprise data platform using data mesh principles β€” multiple business domains (Finance, HR, Operations, Sales) each owning their data products, with centralized governance and shared Fabric capacity.

πŸ—οΈ Architecture Components

All Fabric Workloads Fabric Domains Deployment Pipelines Microsoft Purview Managed VNet Capacity Metrics App

πŸ”„ Data Flow

Sources
SAP, Salesforce, Oracle, SQL Server, APIs, files
β†’
Domain Lakehouses
Each domain runs Bronze→Silver→Gold in its own workspace
β†’
Data Products
Endorsed Gold datasets published as data products with SLAs
β†’
Cross-Domain
Shortcuts + Warehouse views for cross-domain analytics
β†’
Self-Service
Departments build reports from governed data products

πŸ“ Recommended Sizing

Starting SKU F128
Why F128 Multiple concurrent domain workloads + enterprise BI + Spark jobs
Scale trigger F256+ when >5 active domains or >1000 BI consumers

πŸ”’ Governance Model

  • Federated governance: Central CoE defines standards; domain teams implement within guardrails
  • Deployment pipelines: All domains use Devβ†’Testβ†’Prod with branch policies and PR reviews
  • Purview integration: Automated cataloging, lineage, sensitivity labels across all domains
  • Capacity management: Separate capacities for Prod vs Dev; Metrics app monitored by CoE
  • Workspace naming: [Domain]-[Product]-[Env] enforced by policy (e.g., Finance-AR-Prod)
πŸ’°

Chargeback & Cost Attribution

Enterprise Any SKU
β–Ύ

Implement a chargeback model to attribute Fabric capacity costs back to business units, teams, or users. Combines the Fabric Chargeback App with Azure Cost Management for transparent, fair cost allocation.

πŸ—οΈ Architecture Components

Chargeback App Azure Cost Management Fabric Domains Capacity Metrics App Power BI

πŸ”„ Chargeback Flow

Strategy
Define who gets charged, what gets charged, and how to handle edge cases
β†’
Total Cost
Get actual costs from Azure Cost Management (incl. reservations, scale-ups)
β†’
Usage Split
Use Chargeback App to find % usage per domain, workspace, or user
β†’
Bill
Apply cost model: Total Cost Γ— Usage % = Team Charge

πŸ”’ Governance Model

  • Domain enforcement: Every workspace must be assigned to a Fabric domain for accurate attribution
  • Workspace naming: [BU]-[Project]-[Env] convention for easy mapping
  • Monthly review: Chargeback admin validates domain/workspace assignments before bills go out
  • Dispute resolution: Clear escalation path for contested charges
πŸ’‘ How to Use These Templates

These scenarios are starting points, not rigid blueprints. Every customer's needs are unique. Use these as conversation starters:

  • 1. Identify which scenario is closest to your customer's situation
  • 2. Walk through the architecture components and data flow together
  • 3. Adjust the governance model based on their compliance requirements
  • 4. Use the TCO Calculator to estimate costs
  • 5. Start with a pilot using the recommended SKU, then iterate