How decentralising your data with a Data Mesh can improve analytics

Improve profits by decentralising your data with a Data Mesh design.

Data Mesh is the latest buzz word sweeping through the world of data analytics – as a new approach to enterprise wide data architecture.

Jacek Majchrzak – and co-authors Sven Balnojan and Marian Siwiak – have written ‘Data Mesh in Action,’ soon to be published by Manning, but which can be found online.

It explains the benefits of using a decentralised architecture to analyse data to get more insights and increase profitability.

Jacek’s presentation to the Data Vault User Group concentrated on his own experiences of using a data meshand addressed how the new approach can suit businesses of all shapes and sizes.

His thinking – which is expressed strongly in the book – avoids a one-size-fits-all approach. Jacek also explained how he believes you can get more out of your data by adopting domain-driven design principles to create a “data as a product” system.

As an illustration, he began by sketching out how his business had operated 10 years ago, teaching children, aged from six to 17, computer programming, starting out with just one teacher.

Jacek, as the principal, received feedback on the lessons from customers – via a central programming team at the academy, which then made alterations to the curriculum taught to the students.

Issues began as the company expanded with more teachers, and when using a franchise model.

While they provided a template structure for the teaching, the teachers were still given considerable freedom on the content and learning methods.

The problem was the feedback provided gave no context for the data – it was hidden from the decision-makers – so understanding it proved very difficult.

Ideas about changes or modifications to the curriculum being taught came from the top via the academy’s central team.As the business grew and became franchise based, Jacek came up with the idea of a paradigm shift whereby the teachers took ownership of that feedback data.

Instead, by creating a feedback platform – based on simple spreadsheets using Google forms – and getting the franchisees and teachers to pass that feedback to the central team, the process of ownership of the data, its transformation and serving was all decentralised at a stroke.

That increased the value extraction considerably by removing bottlenecks in the relationship between the management, the teacher’s, and the students – and by creating the terminology needed to understand the relationships better.

From there they could create diagrams that identified trends, and that translated the data into business decisions as they developed the curriculum, making teaching better and more informed, as well as increasing customer retention and satisfaction.

And as a result, knowledge and skills were passed to the children, improving business outcomes.

Four principles helped make the new way of working efficient at scale:–

• Domain ownership

• Domain data used as a product

• Federated computational governance

• Utilising a self-serve data platform

In Jacek’s case, the feedback platform helped manage and grow the franchise, increasing profitability.

In the second part of his presentation, Jacek examined the structure of his business, using domain and sub-domain principles, and how the business capability was enhanced thanks to its autonomous and self-governed notion.

By using domain storytelling workshops, businesses can map out their domain to increase understanding of the business’s relationships Jacek also admitted the authors of the ‘Data Mesh in Action’ book had spent a long time trying to come up with a simple definition of what a Data Mesh is before settling on it being the “decentralisation of your data.”

You can listen to some of the questions that were raised about Data Mesh as well as listening to Jacek’s presentation by watching the video above.