Overview of the principles from the DT Reference architecture. The implications have to be added in a later phase

A central design authority governs the data sources and the data integration layers and design and technology decisions on these layers are with this design authority only

Connection and integration of data sources of different types and ability to abstract on the data sources via implementation specific connectors, e.g. for smart meters, wind craft, etc.

Cloud and on-premise deployment options with ability to migrate from cloud to on-premise (cloud is preferred for short term temporary solution)

Data analytics based on data integration layer provides real-time performance and scalability next to batch processing

Data ownership/stewardship and control of the data in the data integration layer. Central management around gathering, integration, and control of data is with the organisation. The single source of truth should include the most possible finest granularity of data.

Data sources can be run on heterogeneous environments, several deployment models (on-premise, cloud, etc.) and can be internal and/or external

Distinguish and decouple classic BI, near real-time BI (Business Intelligence), and real-time OI (Operations Intelligence) and prioritize OI real-time requirements.

Flexibility in terms of integrating different parallel data sources and providing integrated real time data in a scalable way

Flexible customizable analytics functions (self-defined and customizable algorithms) covering rules-based analytics and machine learning

Functional modules and data integration layer can be hosted on different physical environments, e.g. the data integration layer is on-premise while the functionalities are plugged in analytics cloud solutions that work on top of the data

Functionalities can be developed and plugged in on top of dedicated data integration layer as functional modules or can be sourced from 3rd parties (make or buy decision)

Layered architecture that separates data sources, data integration, data analytics, and presentation layers

Prioritized capabilities are incubated in a learning environment to be transitioned effectively in a development, testing, and production environment at a later stage

The functional data analytics layer and the data integration layer are coupled by a standard information model that is binding and under governance of the central design authority

Virtualized infrastructure to allow easy migration in on-premise and also cloud environments