ARTICLES
July 4th, 2022 | 7 min read
Data Mesh: The Next Frontier in Master Data Management?
Data is difficult to manage, especially master data. Master Data Management is essentially the process of controlling the way you define, store, organize and track information related to customers, products and partners across your organization. Traditional data modeling requires that all business entities are defined in advance and cannot change throughout the life of the enterprise system; this limits flexibility and prevents you from easily integrating new systems with existing ones. What if you could handle master data management in a more flexible and agile way? Could it make managing master data much easier?
Data Mesh is a relatively new approach to mastering data, and if its advocates are to believed, could represent the future of master data management.
What is Data Mesh?
Data Mesh is a platform architecture that takes a decentralized approach to managing data. Fundamentally, it’s about treating as something that is separate from any single platform or application. Under the Data Mesh philosophy, data is treated as a democratized product, owned by the entire organisation and managed by domain-specific teams. Each team is responsible for the quality, governance and sovereignty of its own domain, and this data is then provided for use by the rest of the business.
What problems is Data Mesh trying to solve?
The theory behind Data Mesh is a noble one. In essence, it is seeking to solve the fundamental and pervasive issues that have dogged traditional Master Data Management systems for years. In order to realise any value from these initiatives, organisations have been forced to wrestle their data into rigid structures and compositions before they can even get started. Cleaning, preparing and integrating data is hard, time-consuming and expensive, which is why so many Master Data Management projects fail before they've even begun. It's not uncommon in a traditional MDM project for it to take six months to get just one domain operational. Six months! Considering the speed at which customers, markets and competitors change, this timeline simply isn't acceptable. By rejecting the heavily centralized and monolithic models of the past, Data Mesh is trying to do a good thing. The question is whether Data Mesh is a practical alternative.
The challenges of Data Mesh
Despite the buzz around Data Mesh, there are some fundamental questions that need to be answered before you can decide if it's right for you. Firstly, you'll need to consider whether your organisation is operating at a scale at which Data Mesh makes sense. Complete decentralization of data brings its own challenges and risks, as it depends on each of those domain-based teams having the necessary skills and experience to manage the data they are responsible for. Without some level of centralised management, data silos, duplication and governance issues will inevitably arise.
Data Mesh is also a more expensive option. If we accept that every domain-based team will need people with some level of data management experience, and that there needs to be another team to oversee them, suddenly this looks very expensive. For all but the largest of organisations, Data Mesh is most likely cost-prohibitive.
It's not just about cost, it's also about value and building a compelling business case. In theory, federated ownership should lead to quicker learning cycles and accelerated ROI. In practice, if each domain-based team needs to invest in some level of data analytics and engineering expertise before they've even procured any technology, it's going to be very hard for some departments - let’s say the HR team which is responsible for Employee Data - to justify and build a compelling business case.
Which brings us to another point. You cannot buy a Data Mesh. There is no off-the-shelf product that will enable a federated approach to data ownership. It is about so much more than technology and tools. It is a topology, a guiding principle, and in order to realise its value it requires a mindset change and cultural shift that many organisations simply are not ready for.
A means to an end
We've established that Data Mesh is not right for everyone. But that doesn't mean what it seeks to achieve is wrong. Quite the opposite, in fact. There is only way to allow organisations to use their data in ways that will actually help them to adapt to market shifts and serve their customers and stakeholders better. And that is to rip up the current master data management "rulebook" and start again.
It is possible to share data ownership between the IT team and business users without requiring the latter to be data engineers or scientists. It is possible to automate manual tasks like data cleaning, enrichment and integration and save hours of time and significant sums of money. Most importantly, it is possible for sets of data to be treated as products which are universally useful across the business. Truly democratized data can only come from a platform that benefits the many, not only the few - as is the case with Data Mesh. The future therefore lies in a modern approach to Master Data Management that has the same ambition as Data Mesh, but which makes the means of achieving it accessible to all.
