OneLake Shortcuts: Data Without Duplication
Create OneLake shortcuts to access data in ADLS Gen2, Amazon S3, Google Cloud Storage, and other Fabric items — without copying a single byte.
What are OneLake shortcuts?
Think of a shortcut on your desktop.
The shortcut icon points to a file somewhere else on your computer. Double-click the shortcut and it opens the real file. The file itself doesn’t move — you just have a convenient pointer to it.
OneLake shortcuts work the same way for data. You create a pointer in your lakehouse that points to data stored elsewhere — another lakehouse, Azure Data Lake Storage, Amazon S3, or Google Cloud Storage. Your PySpark notebooks and SQL queries read the data as if it’s local, but no data is copied.
This means zero duplication, zero extra storage cost, and always-fresh data.
Shortcut types
| Source | Authentication | Typical Use Case |
|---|---|---|
| Another Fabric lakehouse | Delegated (user identity or workspace identity) | Cross-workspace data sharing without duplication |
| Azure Data Lake Storage Gen2 | Service principal or org identity | Existing data lake → Fabric without migration |
| Amazon S3 | Access key + secret key | Multi-cloud — read AWS data from Fabric |
| Google Cloud Storage | Service account key | Multi-cloud — read GCS data from Fabric |
| Dataverse | Org identity | Power Platform data accessible in Fabric analytics |
| S3-compatible (MinIO, etc.) | Access key + secret key | On-premises or custom S3-API storage |
Where shortcuts can point
- Tables/ section → shortcut appears as a Delta table (queryable via Spark and SQL endpoint)
- Files/ section → shortcut appears as a folder of files (any format)
Creating a shortcut
- Open a lakehouse in Fabric
- Right-click on Tables/ or Files/ → New shortcut
- Choose the source type (Fabric, ADLS, S3, etc.)
- Provide connection details and credentials
- Select the target folder or table
- The shortcut appears instantly — no data copy, no waiting
Scenario: Anika's multi-cloud shortcuts
ShopStream’s data lives in three places:
- Product catalog — existing Azure Data Lake Gen2 (legacy system)
- Payment data — Amazon S3 (payment provider stores data there)
- Marketing events — another Fabric lakehouse in the marketing workspace
Instead of copying data into her lakehouse, Anika creates three shortcuts:
/Tables/ProductCatalog→ ADLS Gen2 shortcut/Files/PaymentRaw/→ S3 shortcut/Tables/MarketingEvents→ Fabric lakehouse shortcut
Her PySpark notebooks join across all three as if they’re local tables. Zero data duplication. Always fresh.
Shortcuts vs mirroring
This comparison appears frequently on the exam.
| Feature | Shortcuts | Mirroring |
|---|---|---|
| Data copied? | No — reads from source at query time | Yes — continuous replication into OneLake as Delta tables |
| Latency | Real-time (reads source directly) | Near real-time (minutes — CDC-based replication) |
| Storage cost | None (no duplication) | OneLake storage for the replicated copy |
| Source types | ADLS, S3, GCS, Dataverse, Fabric items | Azure SQL, Cosmos DB, Snowflake, PostgreSQL, MySQL, Spark catalog |
| Write to source? | No (read-only) | No (read-only replica) |
| Offline access? | No — if source is down, shortcut fails | Yes — replicated data in OneLake survives source outages |
| Best for | Accessing file-based or lake-based data without duplication | Replicating operational databases for analytics without ETL code |
Exam tip: Shortcut vs mirror decision
Use a shortcut when:
- The source is a file store (ADLS, S3, GCS) or another Fabric item
- You want zero data duplication
- The source is always available (no offline access needed)
Use mirroring when:
- The source is a relational database (SQL, Cosmos DB, Snowflake)
- You need a local replica that survives source outages
- You want automatic CDC-based replication without building ETL pipelines
Key exam pattern: “Access data without copying” → Shortcut. “Replicate a database” → Mirroring.
Shortcut considerations
| Consideration | Detail |
|---|---|
| Performance | Reading from external shortcuts (S3, GCS) may be slower than local OneLake data due to network latency |
| Security | Shortcuts inherit lakehouse permissions — but the user must also have access to the source |
| Cost | No OneLake storage cost, but egress charges may apply from AWS or GCP |
| Schema | Shortcut to a Delta table inherits its schema; shortcut to files doesn’t enforce schema |
| Write | Shortcuts are read-only — you cannot write data through a shortcut |
Anika needs to query data stored in Amazon S3 from a Fabric lakehouse. She does not want to copy the data or incur OneLake storage costs. Which feature should she use?
The source behind a lakehouse shortcut (pointing to ADLS Gen2) experiences a 2-hour outage. What impact does this have on Fabric queries that use the shortcut?
🎬 Video coming soon
Next up: Mirroring: Real-Time Database Replication — bring operational databases into Fabric without building a single pipeline.