This document is a generic reference architecture for logical data management.

Logical Data Management Architecture

The logical data fabric  sits in the middle between consuming applications and data sources that can be either intermediate data stores or original data sources such as enterprise applications, operational databases, SaaS, IoT, etc.

It works as a universal semantic layer for consuming applications. Within this layer the user can expose either business domain logical models or Virtual Data Marts for specialized analytics. These semantic models are consumed from BI tools, data science notebooks, AI chatbots, Agentic apps and operational enterprise applications.

Benefits:

• Abstraction / Isolation:
  For consuming applications the origin of data is completely hidden by this data abstraction layer. so data can come from any data source, from an enterprise data warehouse, from a lake house, even from operational enterprise applications or SaaS applications. Consuming applications only depend on these logical semantic models, and are completely isolated and independent of the underlying data models in the data sources.
• Future-proof:
  The user can easily evolve the underlying infrastructure, replacing one system with another avoiding any vendor lock-in, with minimal impact on consumer business applications.
• Universal semantic layer:
  This logical data management approach is universal, holistic, enterprise-wide, so it can integrate any kind of data, from operational enterprise applications, databases, streaming data, cloud applications, SaaS, intermediate analytical stores such as enterprise warehouses, No-SQL data stores or data lakehouses. When deployed at enterprise scale it becomes the universal data foundation for AI including chatbots and agentic applications.
• Bi-directional access to operational systems (R/W):
  The logical data layer goes beyond analytical semantic layers because it can also integrate operational enterprise applications for operational use cases. It can also write back to data sources which is very important for operational usage as well as for autonomous agents that could take actions on their own.

Key Architecture Patterns within a Logical Data Management Architecture

Within a logical data management architecture we can find three key architecture sub patterns:

• Denodo Architectures: Logical Data Warehouse for Analytics
• Denodo Architectures: Data Lake and Lakehouse
• Denodo Architectures: Operational Workloads