July 1, 2023 4 min read

Microsoft Fabric Public Preview: What You Need to Know

Microsoft Fabric Azure Data Engineering Analytics

Microsoft Fabric has been in public preview since Build 2023, and after spending several weeks with it, I want to share a comprehensive overview of what the public preview offers and how you can start exploring it today.

The Preview State

Public preview means the service is available to everyone, but with some caveats:

Features may change before GA
Not recommended for production workloads
SLAs are limited
Some features are still being rolled out

That said, the preview is robust enough for serious evaluation and proof-of-concept work.

What’s Available in Preview

The current preview includes all major Fabric workloads:

Data Engineering

Lakehouses with Delta Lake format
Spark notebooks with Python, Scala, R, and SQL
Spark job definitions for scheduled workloads

Data Factory

Data pipelines for orchestration
Dataflow Gen2 for Power Query-based ETL
Copy activity for data movement

Data Warehouse

T-SQL data warehouse
Cross-database queries
Query insights and performance monitoring

Real-Time Analytics

KQL databases
Eventstreams for streaming ingestion
Real-time dashboards

Data Science

ML models and experiments
MLflow integration
Notebooks with ML libraries

Power BI

Direct Lake mode
Integrated semantic models
Real-time reporting

Preview Limitations

Some features have limitations in preview:

# Spark pool configuration is fixed in preview
# You can't customize these settings yet
spark.conf.get("spark.executor.memory")  # Returns preset value
spark.conf.get("spark.executor.cores")   # Returns preset value

# Workaround: Use session-level configs where supported
spark.conf.set("spark.sql.shuffle.partitions", "200")

Current limitations include:

Fixed Spark pool configurations
Limited regions available
Some connectors not yet available
Capacity limits on trial

Accessing the Preview

To access Fabric preview:

# Check if Fabric is enabled for your tenant
# In Power BI Admin Portal:
# Admin Portal > Tenant Settings > Microsoft Fabric

# For individual trial:
# 1. Go to app.fabric.microsoft.com
# 2. Click "Start trial" if prompted
# 3. Trial provides 64 capacity units for 60 days

Creating Your First Workspace

# Once in Fabric, create a workspace
# Via the UI: Workspaces > New workspace

# Important settings:
# - License: Trial or Premium capacity
# - Default storage: OneLake (automatic)
# - Region: Choose closest available

# Workspace naming convention I recommend:
# {team}-{environment}-{purpose}
# Example: analytics-dev-sales

Preview vs GA Expectations

Based on Microsoft’s announcements, expect these at GA:

More Spark configuration options
Additional connectors
Enhanced security features
Better monitoring and observability
Performance improvements

My Recommendation

If you’re evaluating Fabric:

Start now: The preview is stable enough for learning
Don’t go to production: Wait for GA for production workloads
Test your patterns: Validate your architecture approaches
Document differences: Note any workarounds needed

Sample Evaluation Checklist

Here’s what I’m testing in my evaluation:

# Evaluation areas to test
evaluation_areas = {
    "data_ingestion": [
        "File upload to Lakehouse",
        "API data ingestion",
        "Database replication",
        "Streaming data"
    ],
    "data_transformation": [
        "Spark transformations",
        "Dataflow Gen2",
        "SQL-based transforms",
        "Data quality checks"
    ],
    "data_serving": [
        "SQL endpoint queries",
        "Power BI DirectLake",
        "API access patterns",
        "Cross-workspace queries"
    ],
    "operations": [
        "Monitoring capabilities",
        "Error handling",
        "Scheduling reliability",
        "Cost tracking"
    ]
}

for area, items in evaluation_areas.items():
    print(f"\n{area.upper()}")
    for item in items:
        print(f"  [ ] {item}")

What’s Next

Over the coming weeks, I’ll dive deep into each Fabric workload. Tomorrow, we’ll look at getting started with Fabric - the practical steps to go from zero to your first Lakehouse.