Table layout tool

Lakehouse Table Layout Advisor

Turn table size, file pressure, query filters, update pattern, and engine mix into practical layout and maintenance guidance.

Layout workbench

Use this when a table is getting slower, more expensive, or shared across engines and the layout decision needs evidence.

Directional planning only. Do not paste confidential platform, customer, credential, or incident details.

Assessment

Diagnose table layout from workload signals.

Review file pressure, partitioning, pruning, update pattern, engine mix, and cost evidence before changing layout.

01 · File pressure

Are file counts, file sizes, and small-file patterns visible for production tables?

Small files and uncontrolled ingestion patterns become query and maintenance cost.

Hint

Small files and uncontrolled ingestion patterns become query and maintenance cost.

Best-result move

Review file pressure after new pipelines, streaming jobs, or merge-heavy workloads.

Layout consequence

Choose a level to see what this decision means for the brief.

02 · Partition strategy

Does partitioning match stable, high-selectivity query filters without creating too many partitions?

Partitioning is a contract with future readers and writers.

Hint

Partitioning is a contract with future readers and writers.

Best-result move

Review partition usefulness with observed query and file statistics.

Layout consequence

Choose a level to see what this decision means for the brief.

03 · Clustering and pruning

Are common filters supported by clustering, ordering, or engine-specific pruning features?

Filter-heavy tables need layout choices that make skipping possible.

Hint

Filter-heavy tables need layout choices that make skipping possible.

Best-result move

Refresh clustering choices when consumers or engines change materially.

Layout consequence

Choose a level to see what this decision means for the brief.

04 · Update and merge pattern

Are deletes, merges, late updates, and CDC patterns reflected in maintenance expectations?

High-change tables need different maintenance from append-only facts.

Hint

High-change tables need different maintenance from append-only facts.

Best-result move

Review maintenance cost after CDC volume or late-data behavior changes.

Layout consequence

Choose a level to see what this decision means for the brief.

05 · Engine mix

Do Databricks, Fabric, SQL, BI, or external clients have compatible expectations for this table?

A shared table format does not remove engine-specific behavior.

Hint

A shared table format does not remove engine-specific behavior.

Best-result move

Use engine compatibility checks during release and platform upgrade review.

Layout consequence

Choose a level to see what this decision means for the brief.

06 · Cost evidence

Can the team connect layout choices to scan width, runtime, and maintenance cost?

Layout work should reduce real operating cost, not just satisfy a style preference.

Hint

Layout work should reduce real operating cost, not just satisfy a style preference.

Best-result move

Keep layout decisions tied to cost and performance evidence.

Layout consequence

Choose a level to see what this decision means for the brief.

Checkpoint 1 of 6