Data Mesh and Data Fabric – A Comparison

Mar 11, 2022 |
Robert S. Seiner

President and Principal, KIK Consulting & Educational Services

In the first blog of this three-part series, I explained the overlap and relationship between data governance, data mesh, and data fabric. In this second installment, I will compare data mesh and data fabric and provide observations about the similarities and differences using accountability, methodology, and technology as a basis for judgement.

There are three levels of the basis that must be considered when implementing these types of architectures. The levels are 1) none (there is no accountability or methodology); 2) casual (accountability and methodology are known, but informal and not executed very well); and 3) formal (accountability and methodology are known, active in operations, and there are consequences based on actions). The third level of accountability and methodology, that of being formal and intentional, is recommended to have an effectively governed data environment.

Technology is the glue that pulls together formal accountability and methodology for data mesh and data fabric through the delivery of capabilities that reward the organization with environments and platforms that confidently deliver data to those who will benefit.

Data Mesh

Data Mesh – Accountability

Data mesh architecture provides a behavioral framework for accountability placed as close to the source of the data as possible. The terms “decentralization” and “distribution of accountability” are used to describe the heart of data mesh architecture. The people with distributed accountability are considered business Domain owners (stewards) and their accountabilities are focused on specific domains (or subject areas) of data.

The concept of business domain owner is a positive step toward successful data governance, especially if these people are considered tactical (cross-business area) stewards of their domain. Active governance means that domains must be clearly defined and intentionally managed. The real possibility exists, depending on the distributed nature of your business, that there could be multiple business domain owners for the same subject area of data in different parts of the organization. This is where data governance plays a role.

Effective data governance, which establishes alignment of principles, standards, and practices that ensures data is reliable, consistent and trusted, formalizes an approach to resolve well-intended differences of opinion regarding the definition, production, and use of data. A formal governance program provides proper process to manage, negotiate, compromise, and resolve differences, or escalate the issues/opportunities to a strategic body, often referred to as a data governance council or committee, for resolution.

Data Mesh – Methodology

Decentralization is defined using the words regionalization and federalization. There is data risk associated with regionalization if each region (or part) of the organization can “do their own thing,” business, data, and technology-wise. This is another way that formal data governance plays a role — by providing formal process for applying accountability to build consistent data architecture across the organization.

Many organizations are embracing the concept of federated data governance programs. In the federated model, there is typically a central data governance group or office that provides enterprise standards that must be followed by the distributed entities of the organization. In a federated model, the distributed entities are told what the standards are, but not necessarily told how to follow the standards. In this model, the entities are free to determine their best approach to achieving the standards.

Data Mesh – Technology

Approaches to achieving enterprise standards often require effective use of technology to allow for domain-oriented data governance and analytical capabilities that present a unified view of the organization when necessary. Therefore, even though data mesh focuses on organizational behavior, there is a technology component of data mesh which is addressed by its self-serve data infrastructure principle.

The technical component of data mesh enables delivery of data products that align with data-focused policies through a federated model. Data mesh often includes business domain-based and multi-plane operational and analytical platforms that are managed by dedicated teams. The management includes formalized but decentralized interaction between domains, which often results in enabling data governance as a service (DGaaS) support (the essence of federated data governance) from the central group.

Data Fabric

Data Fabric – Accountability

Accountability is important when it comes to providing a set of services and architecture that deliver reliable capabilities across data environments. Similar to the behavioral aspects of data mesh mentioned earlier in this blog, data fabric requires accountability and standardization of technology data practices across data platforms and devices used to utilize that data.

As mentioned in the previous paragraphs, standardization as a service sounds a lot like DGaaS and requires establishing accountability. In other words, governance, over the behavioral and technical components of data mesh and data fabric architectures. When it comes to simplifying access to and managing data in a heterogeneous environment, data fabric is used to address the behavioral and technical characteristics of the growth of cloud computing and the problems associated with data growth and diversification.1

Data Fabric – Methodology

To reiterate the first blog of the series, data fabric is a set of services and architectures that deliver reliable capabilities across data environments. A data fabric methodology focuses on the use of active metadata to connect disparate environments. The methodology focuses on consistent use of data tools to provide knowledge to the stakeholders, definers, producers, and users (stewards) of the data to enable self-service and provide improved data capabilities.

Data Fabric – Technology

The technical architecture of data fabric delivers effective data access, discovery, integration, security, and lineage capabilities across data environments. Metadata management capabilities are an extremely important component of the architecture required to deliver data democratization competencies. The metadata requires governance and stewardship itself as a resolute effort because the metadata component of the data fabric will not govern itself.

Organizations implementing data fabric architecture aim to deliver a single umbrella of technology that virtually overlays various data repositories while considering the independent requirements of the already distributed and independent tools and systems. The cover (or the fabric) must be governed consistently to assure that there is formal accountability for delivering on required capabilities while executing a proper set of technology standards. The technology of the data fabric, also, will not govern itself.

Summary

As stated earlier, both data mesh and data fabric are connected by their intention to improve the organization’s ability to gain valuable insight and make better decisions from their data. This blog has compared data mesh and data fabric architectures by looking individually at core characteristics of accountability, methodology, and technology, and associating proper data governance and stewardship actions across both.

The final blog of this series will focus on data governance as a directive for both data mesh and data fabric. Organizations taking steps in this direction require a strongly governed environment clearly connected by their intention to improve the organization’s ability to gain valuable insights from trusted data and make better business decisions from their data.

To learn more, watch the on-demand webinar "Does Data Governance Differ in Support of Data Fabric vs. Data Mesh?"