We recently visited Sander Kerstens to talk about his Data Mesh implementation at Vanderlande. Data Mesh is a new approach to organising enterprise data. It aims to make managing and using data easier for everyone involved.
Traditionally speaking, data management is organised through a centralised team that is responsible for all enterprise data. Data Mesh decentralises this, by distributing this responsibility across smaller teams (called domains) within the enterprise. Instead of having central policy and standards that are applied across enterprise teams, teams define their own policies and standards instead.
In Data Mesh, each domain is responsible for the data they generate, the domain decides how their data is managed, processed and shared with other domains. All domains work together in a networked architecture. In turn, allowing for greater collaboration and ability.
A core principle behind the Data Mesh philosophy is one that we often write about: treating data as a product. As a product, there should be clear documentation and standards that describe that data is used and maintained. Much like in traditional data management, Data Mesh stresses the importance of good metadata.
By empowering smaller teams to take ownership of their data and work more closely with other domains, Data Mesh can help organizations to scale and innovate more quickly and efficiently. It bases data management hygiene factors on its principles, rather than having a central data governance team dictate how teams should act.
This introduces a different way of thinking, which may be more suited to modern enterprises. This depends on the culture of the enterprise, though. One approach is not necessarily better than the other, both have their own strengths and weaknesses which are outlined below.
Traditional Data Management
|Greater control and consistency||Potentially slow and inflexible|
|Close alignment with business strategy||May not need team/domain specific requirements|
|Close alignment with regulatory requirements||May not need team/domain specific requirements|
|Agile and responsive to changing business needs||Can help foster innovation and collaboration between teams|
|May present challenges around data quality and consistency||Complex to implement in terms of culture and technical debt|