Data Centricity

A Data-Centric System is a system in which data exists as a first-class entity and is produced, managed, shared, and maintained independently of the applications that create or consume it. Rather than focusing on message exchanges between applications, a Data-Centric architecture focuses on the data itself: its identity, state, structure, relationships, and lifecycle.

In a Data-Centric architecture, data becomes the sole integration contract between producers and consumers. Applications publish data into a shared information space and subscribe to the data they need, without requiring direct knowledge of each other. This paradigm promotes a high degree of decoupling in both space and time.

Spatial decoupling means that data producers and consumers do not need to know each other's location, network address, implementation language, operating system, or even existence.

Temporal decoupling means that communication is inherently asynchronous. Data consumers do not need to be present when data is produced. The infrastructure maintains and distributes information so that applications can access the latest relevant state whenever they become available, including after temporary disconnections, failures, or late joins.

Modelling the Real World

One of the greatest strengths of Data Centricity is its ability to represent real-world entities as they naturally exist:

• Each entity possesses a unique identity.
• Each entity maintains a current state.
• Each entity follows a well-defined lifecycle.
• Each entity has a structure that can evolve over time.

Instead of exchanging transient messages, applications interact through persistent representations of real-world objects and their states.

System-Wide State Awareness

A Data-Centric platform is responsible for maintaining a coherent and consistent representation of the overall system state. Even in the presence of component failures, network interruptions, or application restarts, the infrastructure preserves and reconstructs the latest known valid state.

As a result:

• The current state of the system is always available.
• New or restarted applications can immediately discover and retrieve existing state information.
• Late-joining participants can become operational without requiring special synchronization procedures.
• System recovery becomes significantly simpler and more deterministic.

Multiple Views of Information

Different applications often require different perspectives of the same underlying information.

A Data-Centric system enables applications to create customised views, projections, filters, and subsets of data without impacting the producers of that information. Each application can access exactly the information it requires while remaining independent of how other applications consume the same data.

Capturing Relationships Between Data

Data rarely exists in isolation.

Data-Centricity naturally models relationships, dependencies, ownership, aggregation, and associations between entities. These relationships become part of the information model itself, allowing applications to reason about systems in terms of connected objects rather than isolated messages.

This capability enables the construction of rich digital representations of complex systems such as:

• Air traffic management systems
• Autonomous vehicles
• Industrial automation systems
• Smart cities
• Defense and command-and-control systems
• Large-scale IoT deployments

Beyond State Representation

Although state management is a fundamental capability, Data-Centric middleware extends far beyond simple state distribution.

A Data-Centric platform can efficiently model and support:

Event and Notification-Based Communication

Real-time events, alarms, notifications, and status changes can be represented and distributed with predictable behavior and scalable dissemination.

Request/Reply and Conversational Protocols

Synchronous and asynchronous service interactions can be modelled using data abstractions rather than tightly coupled communication endpoints, enabling efficient implementation of request/reply patterns.

Transactional and Consistent Data Exchange

Data-Centric systems can support lightweight transactional behavior through mechanisms such as coherent sets, grouped updates, and eventually consistent data models, allowing multiple related state changes to be treated as a single logical operation.

Structured and Unstructured Information

The information space can accommodate:

• Well-structured data
• Semi-structured data
• Unstructured data

This allows diverse information sources to coexist within a unified architecture.

Message-Oriented Communication

While Data-Centricity focuses primarily on state and information models, it can also efficiently support traditional message-oriented communication patterns when required by specific subsystems.

The Data-Centric Advantage

Data-Centricity shifts the focus of distributed-system design from application connectivity to information management. By elevating data to the primary architectural element, organizations gain:

• Loose coupling between applications
• Greater scalability
• Improved resilience and fault tolerance
• Simplified system evolution
• Stronger interoperability
• Natural representation of real-world entities
• Continuous visibility of system state
• Faster integration of new applications and services

The result is an architecture that is inherently more adaptable, maintainable, and resilient, particularly for large-scale, real-time, mission-critical, and long-lived distributed systems.