DataMosaic documentation

How to set up DataMosaic, build a Check that compares two systems, and read what it finds.

Two systems are asked the same question and give two answers. The till says 91. Finance says 86. Somebody has to work out why, and they have to do it before the month closes.

DataMosaic — Mosaic from here on — is automated data verification for retail IT. It reads from the systems you already run, compares them on the keys and measures you choose, and tells you where they disagree and by how much.

#Start here

If you have never opened Mosaic before, read these four in order. They take about twenty minutes end to end and finish with a Check that has run.

#How this documentation is arranged

Getting started is the guided path above: read it front to back.

Concepts explains each idea on its own — what a data source is, what normalising does, how a Check differs from a Report. Read a page when you meet the idea, not before.

Reference sections describe one building block per page, always in the same order: what it is, when to use it, every field it offers, and what commonly goes wrong. You are meant to arrive at these from a link or the search box rather than read them through.

#Three words worth knowing now

The rest of the documentation leans on these, and two of them are easy to mix up.

• A Check compares two or more systems and reports where they differ. It is the thing you build.
• Rules are the pass-or-fail thresholds on a Check — a difference of more than two rows is a problem. A Check with no Rules still runs and still shows differences; it simply never grades itself.
• A Report is the flat kind: one source, shown as it comes. It compares nothing, so it has no Rules.

There is a fuller glossary once the ideas start overlapping.