Vpax Analysis

Back

Model Storage & Performance Report

Analyze

Overview

Features

Prompts

Help

Key Features

A VPAX export from DAX Studio is full of useful numbers — column sizes, row counts, cardinality, encoding types — but it never tells you whether any of them is actually a problem. Is a 40-million-row column fine, or a red flag? Should that column be hash-encoded or value-encoded? Working that out means knowing the thresholds by heart and reading the numbers one by one. This skill does the reading for you: it runs 30 checks across model size, cardinality, encoding, relationships, column design, compression, and measures, ranks what it finds by how much it actually hurts, and points you straight at the tables and columns worth fixing first — run it with /vpax-analysis.

How it works

Before you start — this skill reads a VPAX file, so you make that in DAX Studio first: open DAX Studio, connect to your model, click View Metrics on the Advanced tab, then click Export Metrics and save a .vpax (full steps in Setup below). Put that file in inputs/vertipaq/, or just tell Claude where you saved it and it'll move it for you.

Then type /vpax-analysis, point it to your VPAX file, and the skill does the interpretation for you:

Extracts model metadata from the VPAX archive
Evaluates 30 rules across 7 categories: model size, cardinality, encoding efficiency, relationships, column optimization, compression, and measures
Flags violations by severity with impact tags (Storage, Query, Refresh, Data Quality, Maintainability)
Builds supplementary data: top tables by size, top 20 columns by size with cardinality and encoding

What you get