🛡️Governance & Quality
Keeping data accurate, documented and trustworthy · 41 terms
Data Governance
10The written rules for data, inside a firm, in EU law and in Europe's shared rule books, the people who own and run it, the skill of reading it, and the shared lists and records every team uses.
The rules and roles for data. They say who owns it and who may use it.
A written rule for data. It says who may use it and how long to keep it.
An EU law that lets users get the data their devices make. It also eases cloud switching.
A European project that writes rules and labels for cloud and data. It runs no cloud.
The skill to read, use and talk about data. All of us need some.
One true copy of each key thing, like a customer.
A short shared list of allowed codes, like country codes. It rarely changes.
Governance Roles
3The people in governance: who decides about a data set, who keeps it clean and who runs the systems that hold it.
The person in charge of one data set. They decide who uses it and how.
The one who keeps a data set clean. They do it each day.
The team that runs the place the data lives, and keeps it safe.
Data Quality
17How good data is, the ways to judge it, the checks that test it and the tools that watch it.
How fit data is for its use. It should be right, complete and on time.
Quality Dimensions
7The separate ways to judge data: is it right, full, in agreement, fresh, free of copies and within the rules.
The ways to judge data. Is it right? Is it fresh?
Whether a value matches the real thing it describes.
Whether all the rows and fields that should be there are there.
Whether the same fact matches in every place it is kept.
How up to date data is, measured as the time since its last load.
Whether each thing is in the data just once.
Whether a value keeps to the rules, like a real date.
Quality Checks
5Looking at what data holds, the rules it must pass in the pipeline, the tools that run them, and checks that copies still match.
Counting what is really in a table: blanks, values and ranges. It comes before checks.
A rule data must pass. The pipeline tests it each time.
A free Python tool that tests data. It comes with lots of checks.
A written deal on what data will look like. Its maker keeps to it.
Checking that copied data still matches its source. You compare counts and totals.
Data Observability
4Watching data all the time for odd days, changed columns and late loads.
Watching data all the time, and calling out when it looks odd.
Time when data is missing, late or wrong. It counts from the break, not the alert.
When the columns of new data change and no one says so.
A promise of when data will be ready, and how good.
Metadata
8Data about data: the catalog that finds it, the words and columns that explain it, and the map of where it came from.
Data about data. Who owns this table? When did it last change?
A search box for all of a firm's data. Each table gets a page.
A free data catalog you can run yourself. It was first built at LinkedIn.
Tools to find, mark and guard data, made by Microsoft. One part is a catalog.
A paid tool for data rules. It keeps the catalog, the glossary and who signs off.
A shared list of what a company's key words mean.
A list of each column in a table and what it means.
A map of where data came from and where it goes next.
Access Control
3Who may read or change data, set with grants and roles, down to single rows and columns.
Rules for who may read or change each table, row and column.
One table, with rows hidden by person. Each reader sees only their own rows.
People may read a table but not all its columns. The private columns stay closed.
Data Mesh
3Splitting data work by business area, where each area owns its data and shares it as a product.
Each team owns its own data and shares it with the rest.
Data made to be shared, with an owner and clear promises.
One part of the business, like orders. It owns its own data.