Uniting data mesh and data fabric within a single, modernized architecture

Modernization techniques continue to be a hot topic in adapting to an abundantly data-minded world, where topics like data fabric, data mesh, data-centric, and FAIR often arise; with varying techniques and technologies surfacing, identifying what your organization specifically needs to succeed is becoming a difficult task. 

Sean Martin, CTO and co-founder at Cambridge Semantics joined KMWorld’s webinar, “Data Fabric or Data Mesh? What’s the difference?” to explore the nuances between these trending digital transformation strategies and methodologies, explaining that knowledge graph technology is the critical component to embracing a successful and agile data architecture.

Martin began the conversation by focusing on the basics: the difference between a data mesh and a data fabric:

• Data mesh is a bottom-up approach that decentralizes data responsibility and centralized infrastructure, such as data warehouses.
• Data fabric is a top-down approach that integrates data across an organization, devolving responsibilities for datasets closer to where the data is produced.

Importantly, Martin noted that both approaches can be complementary to one another, working in tandem to achieve a truly modern data architecture. The key to their unification lies in semantic knowledge graph technologies, according to Martin.

While both data mesh and data fabric rely on creating reusable data products, semantic technologies represent the ideal tool for crafting these data products. Semantic technologies, including a semantic knowledge graph, accommodate for complex relationships and descriptions which lend itself to connecting data with context—which is particularly optimal for implementing data fabrics due to integrating a myriad of data sources, data types, and schema.

In combination with semantic technologies’ ability to connect and give context to data, successfully merging data mesh and data fabric requires a two-tiered architecture, Martin pointed out.

The first tier—the data mesh—provisions and describes data, while the second tier—data fabric—integrates data products across domains. The semantic technologies in place ensure that any of the datasets are readily understood by business users by utilizing business-friendly terminology, affording organizations the advantages of each architecture in one, comprehensive framework.

Martin further explained that AI can be useful in embracing a merged, data mesh-data fabric architecture—though it's not without its limitations. Martin argued that AI’s capabilities for data integration have been exaggerated, as it has limited functionality in integrating data for data fabric without human effort.

However, AI certainly has a role to play in this sort of digital transformation: automating the creation of knowledge graphs which serve as the truth backbone to a data fabric and data mesh. Martin explained that there are several AI techniques for identifying connections in datasets, which will best prepare data mesh and data fabric for success.

Martin described the material benefits that a merged data mesh-data fabric architecture can introduce to any business. These advantages include:

• Reduced ETL and ETL processing
• Decreased cost
• Reduced time-to–value
• Enhanced analytics and value derived from data
• Better, more comprehensive understanding of data relationships

For an in-depth review of data mesh and data fabric architectures, you can view an archived version of the webinar here.