Data Analytics trends & challenges
The 16th MIT Chief Data Officer and Information Quality Symposium (CDOIQ) Conference this year brought together some 2000 data leaders from 60+ countries — thanks to the conference organizers.
- Data Mesh vs Data Fabric: was the battle royale of the conference. The clash of the two methodologies came up in almost very presentation. The battle seems to have heated up after Gartner recently moved Data Mesh to Obsolete category in their latest Hype Cycle. I am quoting below some comparisons that were mentioned during the conference:
“Data mesh is a highly decentralized data architecture focused on addressing challenges such as lack of ownership of data, scaling bottlenecks and lack of quality data. You can think of Data Mesh as treating Data as a Product whereby each source has a data product owner which could be part of a cross functional team of data engineers. Data Fabric on the other hand, is an all in one integrated architectural layer that connects data and analytical processes. It integrates data, analytics and dashboards into one solution that helps accelerate access to data in a distributed environment” — Violet Wittig
“Data Fabric and Data Mesh are very different, but are both reference architectures. The former assumes that AI and metadata will allow for data to inform its own classification and use/governance. A data fabric does not require centralized data persistence, but does need a common data management / semantics layer. The data mesh is a hybrid of decentralized and centralized data management that gives governance over data to individual business ‘domains’, but where shared master data is still managed at a centralized level. Fabrics are highly conceptual and several years from main stream, whereas a data mesh could conceivably be implemented now for companies with higher data governance maturity.” — Malcolm Hawker
“Data Mesh tries to argument that data should be packaged as a product by those who create it. Data Mesh products are packets directly from business operations without necessarily knowing how data will be used by second data consumers. Data Mesh is data supplier approach in TDQM terms. Mesh Product could be clinical laboratory invoices. Data fabric is about…