Nothing has galvanized the data community more in recent months than two new architectural paradigms for managing enterprise data. On one side there is the data fabric: a centralized architecture that runs a variety of analytic services and applications on top of a layer of universal connectivity. On the other side, is a data mesh: a decentralized architecture that empowers domain owners to manage their own data according to enterprise standards and make it available to peers as they desire.

Most data leaders are still trying to ferret out the implications of both approaches for their own data environments. One of those is Srinivasan Sankar, the enterprise data & analytics leader at Hanover Insurance Group. In this wide-ranging, back-and-forth discussion, Sankar and Eckerson explore the suitability of the data mesh for Hanover, how the Data Fabric might support a Data Mesh, whether a Data Mesh obviates the need for a data warehouse, and practical steps Hanover might to take implement a Data Mesh built on top of a Data Fabric.

Key Takeaways:

- What is the essence of a data mesh?

- How does it relate to the data fabric?

- Does the data mesh require a cultural transformation?

- Does the data mesh obviate the need for a data warehouse?

- How does data architecture as a service fit with the data mesh?

- What is the best way to roll out a data mesh?

- What's the role of a data catalog?

- What is a suitable roadmap for full implementation?