You are Home   »   News   »   View Article

How Equinor implemented the ‘data mesh’ concept

Wednesday, November 29, 2023

The 'data mesh' concept is about 'packaging' a company's data management, so the output of every company department becomes a 'product' for any department which uses it. Equinor is trying out the idea

Equinor was seeing a great deal of 'challenges and friction' when trying to use its data across different company departments, said Nina Reiersgaard, Advisor Enterprise Data Management, Equinor, speaking at the Society for Professional Data Managers conference in November.

The company was interested in ideas for how organisational data could be better organised.

Company staff liked the corporate data management philosophy described in a Harvard Business Review article 'Democratising Transformation', which has Microsoft's CEO Satya Nadella as a co-writer. It describes the practise of different divisions of the company making data 'platforms' with associated governance and documentation.

The company was also 'inspired' by a Harvard Business Review article, 'A better way to put your data to work" by McKinsey consultants Veeral Desai, Tim Fountaine and Kayvaun Rowshankish. This has the idea that data is seen as a 'product', as it is made available to other parts of the company, like products on a supermarket shelf.

This then makes it easier for other employees to discover and use data which might help them. The data can also be made available to machines, for use in machine learning projects.

In this model, the company should have some centralised governance about how data is managed, just as any company has some level of centralised governance. This ensures that each department provides high quality data.

The aim is to fix the biggest challenges employees have with data, according to employee surveys. Employees regularly say that they don't know what data the company has and they can't find it easily; they don't have the skills to work with data; and they don't know which data is trustworthy.

The initiative can address all three of these challenges, Ms Reiersgaard said, by making it easier to find and access data, which is already prepared to be easy to work with, and trustable.

The initiative should also help the company better allocate its training budget, so that it is perhaps less orientated around specific software packages. Equinor has done some analysis of its spending on digital competency, and found that sometimes only 20 per cent of staff are using a certain tool, a year after they have completed a training course on it.

The data mesh approach can help remove many [work] bottlenecks. Rather than one team being responsible for too much, work is more distributed, she said. 'That makes a lot of things go faster compared to how it was.'

For data to be considered a 'product', it needs to have an 'owner' who is responsible for it, and a clear understanding about the data quality and the rights to use it, she said. It should also have context around the data, so people know what it is.

'We are implementing and publishing data products with the relevant context in our data catalogues,' she said.

Equinor is starting its 'data product' re-organisation by starting with the data which has the highest value, or which is most used in the company, she said. There is a particular focus on 'master' or 'reference' data, which is widely used and relied upon.

The work is taking place in 'sprints'. There are 7 teams, who all join a review meeting every second week, to show what they have done. The work is happening iteratively, 'not everything will be solved at once,' she said.

Altogether there are 60-70 people involved in these 'enabling activities', working together with domain specialists,' she said.

The OSDU data framework, which Equinor had a big role in setting up, will need to fit into this mesh system. The company is still working out how to do that, she said.

Federated governance

The company's 'federated' data governance model means that certain aspects of data governance are decided for the company as a whole. This includes decisions about which data standards will be used in the company.

It has set up a 'data governance council' providing central co-ordination and ensuring that the governance systems are 'relevant across all business areas.'

The company has many data teams distributed around the company. The common governance ensures they follow the same standards, she said.

The company is divided up into both business 'areas' (such as for a certain part of the world or a certain asset), and 'functional domains' such as reservoir. Some functional domains have responsibilities for multiple business areas, such as a reservoir team being responsible for reservoirs in different parts of the world.

In the past, each business area has had its own IT department. The IT departments are being more joined up, so that there can be a closer connection between data product owners, data stewards and technical data delivery teams, she said.

Reducing data engineering

The company is keen to find ways to reduce the amount of data engineering / integration work.

At the moment, data engineering is needed for every data task, and maybe several iterations of it, she said. Perhaps some of this data engineering effort could be re-used.

For example, a typical data engineering task might be connecting together data from safety or maintenance management, sometimes by developing mapping tables. It is a similar picture integrating data about pieces of equipment, wells and subsurface. It may be possible to re-use these mapping tables, for every time these data sources are used together.

Associated Companies
» Equinor
comments powered by Disqus


To attend our free events, receive our newsletter, and receive the free colour Digital Energy Journal.


Latest Edition Mar-Apr 2024
Apr 2024

Download latest and back issues


Learn more about supporting Digital Energy Journal