Data Mesh — How to Make Data Governance “Computational”

How to write policies that are automatable, implement them and grow your rate of computational governance as your company matures

(Photo by David Levêque on Unsplash)

In this article, I’ll explain a simple framework that will help you working on the “computational” parts of your data governance inside a Data Mesh.

We will:

• Explain what “computational” means, but do not cover “federated computational governance” in all details. For that you can check out Zhamaks original article on the subject.
• We will then talk about two necessary steps for making governance policies entirely computational.

This article is an extract from a draft of the book “Data Mesh in Action” in MEAP at Manning, which I am co-authoring.

Computational Data Governance means automated data governance.

With computational data governance, we refer to automated governance. That in turn means we have some way of automatically checking some data policies.

The key to roling out computational governance is to understand, that we implement it from the very beginning. We just start with very small and easy to implement policies. We then later roll out more advanced policies as we iterate.

The three necessary steps for making governance policies entirely computational are:

• Step 1 is to bring the policy in a form that can be automated,
• Step 2 is to automate the execution,
• and Step 3 is to increase the rate of automation.

You can choose a level of automation that fits your current level of data maturity and slowly increase that by retaking the first two steps from time to time, following the framework we show in step 2 & 3.

Our starting point is an understanding of our policies, which we should now have in a written form. They could read, for instance:

List of Policies 1. No personalized information is allowed to be shared without strict access control.

2. Every data should contain metadata that identifies…