Fabric Architecture

Fabric Lakehouse

# Fabric Lakehouse with PySpark
spark.read.format("delta").load("abfss://onelake@fabric-fabric.pbidedicated.microsoft.com/sales.lakehouse/Tables/raw_sales") \
    .filter("amount > 0") \
    .write.format("delta").mode("overwrite").save("abfss://onelake@fabric-fabric.pbidedicated.microsoft.com/sales.lakehouse/Tables/curated_sales")

# Fabric SQL Analytics Endpoint
# Connect via: fabric.sql.azuresynapse.net
# Use standard T-SQL queries

Fabric Data Factory

{
  "name": "fabric-pipeline",
  "type": "DataFactory",
  "properties": {
    "activities": [
      {
        "name": "CopyData",
        "type": "Copy",
        "inputs": [
          { "referenceName": "adls_source", "type": "DatasetReference" }
        ],
        "outputs": [
          { "referenceName": "lakehouse_sink", "type": "DatasetReference" }
        ]
      }
    ]
  }
}

Interview Questions

Q1: How does Fabric differ from Synapse Analytics? A: Fabric is a SaaS platform with unified storage (OneLake), compute engines, and Power BI. Synapse is a PaaS service with dedicated/serverless pools. Fabric provides simpler management and tighter integration.

Q2: What is the Lakehouse pattern in Fabric? A: Lakehouse combines data lake storage with data warehouse capabilities. In Fabric, Lakehouse uses Spark compute to query Delta tables stored in OneLake, providing SQL and programmatic access.

Q3: When would you use Fabric vs Azure Data Engineering services? A: Fabric for organizations wanting unified SaaS analytics with minimal infrastructure management. Azure services for teams needing more control, flexibility, or specific service capabilities.