In Microsoft Fabric, performance is everything. While V-order, Z-order, and Liquid Clustering all aim to make your queries faster, they work at different levels of the data “stack.”
V-Order
The “Inside” Optimizer.
V-order is a Microsoft-proprietary optimization that happens inside each Parquet file. It doesn’t decide which rows go into which file; instead, it rearranges data within the file to maximize compression and read speed for the VertiPaq engine (used by Power BI Direct Lake).
- Level: File-level (Internal).
- When: Applied during write-time (by default in Fabric).
When: Applied during write-time (by default in Fabric).
Z-Order
The “Classic” Co-locator.
Z-order is a standard Delta Lake technique. It rearranges data across files so that related information is stored together. For example, if you frequently filter by CustomerID and Date, Z-order will group those rows into the same files, allowing the engine to skip files that don’t match your filter (Data Skipping).
- Level: Table-level (External/File Layout).
- Constraint: If you want to change the columns you Z-order by, you must rewrite the entire table.
Liquid Clustering
The “Future” of Layout
Liquid Clustering is the successor to Z-order and Partitioning. It is incremental and flexible. It uses a tree-based algorithm to cluster data as it arrives. Unlike Z-order, you can change your clustering keys without a full data rewrite.
- Level: Table-level (Modern Replacement for Z-order).
- Benefit: Handles data skew automatically and works great for tables that grow rapidly.
| Feature | V-Order | Z-Order | Liquid Clustering |
| Focus | Compression inside the file | Data skipping across files | Dynamic/Flexible data skipping |
| Compatibility | Works with Z or Liquid | Works with V-order | Works with V-order |
| Flexibility | Automatic | Hard (requires full rewrite) | High (easy to change keys) |
| Best For | Power BI / Direct Lake speed | Static query patterns | Changing patterns / Rapid growth |
Ultra-short memory trick
V-Order optimizes column layout for BI, Z-Order optimizes filtering, and Liquid Clustering adapts data layout over time.
- V-Order = BI scan
- Z-Order = WHERE filter
- Liquid Clustering = Future-proof layout
The Best Setup for 2026
In a modern Fabric project, you should almost never use Z-order anymore. Instead, use the Golden Combo:
- V-Order (Internal): Keep this enabled to ensure Power BI is lightning fast.
- Liquid Clustering (External): Use
CLUSTER BY (ColumnName)on your large tables to handle file-level efficiency.