Architecture paradigm where data responsibility is decentralised to domain teams and data is treated as a product with clear ownership.
Definition
Data Mesh is a data architecture strategy introduced by Zhamak Dehghani (2019) as an alternative to centralised data warehouses and data lakes. Rather than a single central data team owning and managing all data, responsibility is distributed to the domain teams that produce the data - treating data as products with defined APIs, quality guarantees and discoverability. Four principles: domain-oriented ownership, data as a product, self-serve data infrastructure and federated data governance. Data Mesh is particularly relevant for large organisations with many systems and data silos - such as grid companies with SCADASupervisory Control and Data Acquisition - system for real-time monitoring and control of industrial processes., MDMMeter Data Management - system for collection, storage, quality assurance and distribution of metering data from the electricity grid., CIS, GISGeographic Information System - system for collecting, storing and analyzing location-based data - used in network mapping and planning. and ERPEnterprise Resource Planning - business system integrating finance, HR, logistics and operations., all with different data models and ownership. Data Mesh does not solve technical problems, but organisational ones: who is responsible for ensuring data is correct and available?
In practice in the Norwegian energy sector: a grid company with Data Mesh will have an MDM team that owns and exposes 'validated hourly values per metering point' as a data product - with an SLA covering update frequency, format and quality. The CIS team consumes this product rather than fetching data via ad hoc ETL jobs. For CSRD reporting, Data Mesh means that energy data (Scope 2 consumption) can be retrieved from a defined data product with documented data quality - rather than from manual Excel extracts. Data Mesh is a mature concept and requires organisational maturity - it is not a technical shortcut.
In practice
In practice in the Norwegian energy sector: a grid companyCompany that owns and operates electricity grids and is responsible for power supply to end users. with Data Mesh will have an MDMMeter Data Management - system for collection, storage, quality assurance and distribution of metering data from the electricity grid. team that owns and exposes 'validated hourly values per metering pointThe fundamental unit in Elhub - the unique point in the electricity network where energy is measured and settled, identified by an 18-digit GSRN number.' as a data product - with an SLAService Level Agreement - contract defining guaranteed availability and performance for a service. covering update frequency, format and quality. The CIS team consumes this product rather than fetching data via ad hoc ETL jobs. For CSRDCorporate Sustainability Reporting Directive - EU directive requiring detailed sustainability reporting from large companies starting 2024. reporting, Data Mesh means that energy dataCollective term for all forms of digital data related to the production, distribution, and consumption of energy. (Scope 2 consumption) can be retrieved from a defined data product with documented data qualityThe degree of accuracy, completeness and reliability in a dataset - critical for correct billing and balance settlement in the power sector. - rather than from manual Excel extracts. Data Mesh is a mature concept and requires organisational maturity - it is not a technical shortcut.
