One of the earliest indicators of a discipline’s level of maturity is the creation of different strategies, often even in apparent opposition, which on closer analysis turn out to be different tools suitable for different contexts.

In the case of data management, among the many modes and strategies, at least three organizational principles can be identified: data lake, data fabric and data mesh. Precisely the latter, on which we will focus, offers some peculiar features because, unlike the custom, it starts from the idea of applying a decentralization of data, reinforcing the concept of ownership and responsibility for each organizational unit.

What is the Data Mesh and how does it work?

The concept of data mesh was introduced by Zhamak Dehghani¹ in 2019. This paradigm was born as a response to the limitations of more canonical solutions such as data warehouses and data lakes.
The main difference is that where hitherto traditional approaches were based on the idea of centralizing data, the data mesh principle promotes decentralized management, in which each domain, which we might liken to an organizational unit, is responsible for its own data.

This decentralization overcomes some of the limitations of more formal, monolithic approaches, for example, poor scalability, management complexity, and poor data quality.

The four cardinal principles

The data mesh, in the original form proposed by Zhamak Dehghani, is based on four basic principles, which we recall:

  1. Domain-oriented data ownership and management: in a data mesh, data are organized around business domains, each of which is responsible for its own data and its management.
  2. Data as product: data must be managed as products, and as such organized, maintained and made accessible.
  3. Self-service data infrastructure on shared platform: starting from common tools and infrastructure, each domain can manage data independently.
  4. Federated governance: ensures common guidelines and standards.

The last point in particular suggests that the decentralization proposed by the data mesh should not be confused with the reconstruction of corporate silos or disorganized data management. On the contrary, precisely in order to guarantee each unit a high level of autonomy , it is essential that the tools, platforms and guidelines are rigorously managed and organized.

When to use the data mesh? The benefits

Now it is clear that data mesh architecture has some peculiarities that make it particularly suitable for certain contexts. To clarify when it can be considered the ideal approach, let us recall its main advantages.

Let’s start with scalability and flexibility: each federated organizational unit can independently manage its own solutions, even choosing them from a size perspective. In this way, the response to needs will be faster and more appropriate. In addition, data quality is typically very high precisely because each domain manages and controls its own information.

Clarity of responsibilities and greater contextual knowledge are a guarantee of better management work. The data mesh approach is also attractive from the point of view ofagility: since different units have a high degree of independence, they can act more quickly to meet their needs. For the same reason, this mode also guarantees a reduction in bottlenecks, since operations do not have to be centralized.

Data Mesh Technology: application realities

Like any tool and architecture with strong distinctive features, data mesh technology, while having considerable advantages, is not suitable for all contexts and all types of companies. In general terms, precisely because of the independence it grants to organizational units, it offers the best results in realities that have a high level of digital maturity. Here are some examples:

Complex organizations -Companies with complex organizational structures, with many departments and operating units, can get the most out of the decentralization offered by data mesh. Each unit operates independently, with fewer conflicts and areas of friction with others.

High volume of data: organizations that produce and must manage large volumes of data often find it difficult to harmonize everything in one centralized system. A data mesh makes it possible to distribute the workload by segmenting management into simpler units.

Highly dynamic need: in industries where rapid changes and sudden responses are required, data mesh offers the flexibility to innovate and adapt quickly.

Data Mesh: independence, agility, orchestration

Data mesh undoubtedly represents an innovative approach based on decentralization at a time in history when convergence dominated the market. With its principles of domain independence and federated governance, the data mesh architecture is a powerful and versatile solution for companies wishing to tackle complexity with an innovative approach.

Do you want to learn more? Tell us your needs, our experts are ready to provide the answers you seek.

¹Source: Jasse-Anderson.com