Data architecture is how organizations organize and manage their data. Different types of architecture solve various data challenges. Here are the main types of data architecture used today.

Centralized Data Architecture

In a centralized data architecture, all data is stored and managed in one central location. Users and applications connect to this central system to access or modify the data. This central system handles all data requests, ensuring that information remains consistent and secure throughout the organization.

Pros: Easy to control and secure data. All data is in one place, making it simple to find.
Cons: If the central system fails, all data access is lost. It can be slow for users far from the central location.

Data Flow in Centralized Data Architecture

Data is collected from various sources.
It is stored in the central system.
Users request data from this central point.
The central system processes these requests and returns the data.

Decentralized Data Architecture

Decentralized data architecture spreads data across multiple independent locations or systems. Each location manages its own data without depending on a central authority. Users access data from the nearest or most relevant location, and sharing data between sites may involve additional steps or agreements.

Pros: Faster access for local users, and if one part fails, the others can still function.
Cons: It can be more challenging to keep data consistent across all locations, and overall management may become more complex.

Data Flow in Decentralized Data Architecture

Data is created and stored locally.
Users access data from their nearest or most relevant location.
Updates happen locally, without need for central coordination.
Sharing data between locations may require additional processes.

Distributed Data Architecture

Distributed data architecture spreads data across multiple connected systems or nodes. These nodes work together as part of a single system, sharing the workload. Each node can process requests independently, but they communicate with each other to share data when needed. A central system may oversee the distribution and access of data across all nodes.

Pros: Offers a good balance of speed and consistency, and can handle a large number of users and data effectively.
Cons: Can be complex to set up and manage, and requires reliable network connections.

Data Flow in Distributed Data Architecture

Data is distributed across various connected nodes.
Each node can process requests independently.
Nodes communicate with each other to share data as needed.
A central system may oversee the distribution and access of data.

Data Warehouse Architecture

Data Warehouse architecture collects and stores data from different sources in a structured format designed for analysis. It extracts data from different systems, transforms it to fit a standard structure, and loads it into the warehouse. Users can then query this aggregated data for reporting, analysis, and decision-making purposes.

Pros: Useful for overall analysis and keeps historical data.
Cons: Can be costly, and the data might not always be up to date.

Data Flow in Data Warehouse Architecture

Data is extracted from various source systems.
Data is transformed to fit the warehouse structure.
Transformed data is loaded into the warehouse.
Users query the warehouse for analysis and reporting.

Data Lake Architecture

Data Lake architecture stores large amounts of raw data in its original format, accepting all types of data without the need for pre-processing. When users want to analyze the data, they can access it directly from the lake and process it as needed. This allows for flexible and diverse types of analysis on the same data set.

Pros: Can store any type of data and is flexible for various uses.
Cons: Can become disorganized if not managed properly, making it difficult to find specific data.

Data Flow in Data Lake Architecture

Raw data is collected from various sources.
The data is stored in its original format.
Users can access and analyze the data as needed.
Processing and structuring take place at the time of use.

Cloud-Based Data Architecture

Cloud-based architecture uses remote servers accessed over the internet to store, manage, and process data. It allows organizations to use computing resources on demand without maintaining physical infrastructure. Users can access data and services from anywhere with an internet connection, and the system can easily scale resources up or down based on needs.

Pros: Can be accessed from anywhere and is easy to scale up or down
Cons: It depends on a stable internet connection, and there may be security concerns to consider.

Data Flow in Cloud-based Architecture

Data is uploaded to cloud storage.
Cloud services process and manage the data.
Users access the data through web interfaces or APIs.
Resources automatically scale up or down based on demand.

Edge Computing Architecture

Edge Computing architecture processes data close to its source, such as on devices or local servers, rather than sending it to a centralized system first. This enables faster processing of time-sensitive data. Only relevant data or results are then sent to the central system for long-term storage or further analysis.

Pros: Offers faster response times and reduces the amount of data sent over networks
Cons: Limited processing power on edge devices and can be complex to manage.

Data Flow in Edge Computing Architecture

Data is generated by devices, such as sensors and IoT devices.
Immediate processing takes place on nearby edge devices.
Only relevant data or results are sent to the central system.
The central system manages long-term storage and further analysis.

Microservices Architecture

Microservices architecture breaks down an application into small, independent services with each service responsible for a specific function and managing its own data. These services communicate with each other through well-defined interfaces or APIs. This allows different parts of the system to be developed, deployed, and scaled independently.

Pros: Flexible and easy to update, with different parts able to use various technologies.
Cons: Can be complex to manage all the components and might face data consistency issues.

Data Flow in Microservices Architecture

Data is generated by devices, such as sensors and IoT devices.
Immediate processing takes place on nearby edge devices.
Only relevant data or results are sent to the central system.
The central system manages long-term storage and further analysis.

Lambda Architecture

Lambda architecture processes data using two parallel systems: a batch layer for handling large amounts of historical data, and a speed layer for processing real-time data. A serving layer then combines results from both layers to provide comprehensive views of the data. This allows the system to handle both high-volume batch processing and low-latency real-time data analysis.

Pros: Handles both real-time and historical data, offers low-latency reads and updates, and provides a comprehensive view of the data.
Cons: Can be complex to implement, may lead to data inconsistency, and requires managing two separate systems.

Data Flow in Lambda Architecture

Data enters the system and is sent to both the batch and speed layers.
Batch layer processes historical data.
Speed layer processes real-time data.
Serving layer combines results for queries.

Kappa Architecture

Kappa Architecture is a simpler version of Lambda Architecture that treats all data as a stream. It uses a single stream processing system for both real-time data and historical data reprocessing, eliminating the need for separate batch and speed layers. This approach reduces complexity by using the same code and infrastructure for both data types.

Pros: Simpler than Lambda, provides consistent processing for all data, and is easier to maintain and update.
Cons: Less efficient for large batches, needs a robust stream processing system, and is limited to suitable use cases.

Data Flow in Kappa Architecture

All data enters the system as a stream.
The stream processor handles both real-time and historical data.
Processed results are stored in a serving layer.
Queries are served from the serving layer.
To reprocess data, the stream is replayed from the start.

Event-driven Architecture

Event-driven Architecture focuses on the production, detection, and response to events. An event is any important change, like a user action or a sensor reading. Components communicate by sending and receiving events. When an event happens, the system quickly processes it and takes the necessary actions, often triggering new events in response.

Pros: Highly responsive; Loosely coupled components; Scales well for real-time processing
Cons: Complex design and debugging; Challenges with event ordering; Potential for event storms

Data FLow in Event-driven Architecture

An event occurs and is published to the event channel.
Event consumers subscribe to the relevant channels.
Those consumers receive the published events.
Consumers then process the events, which may trigger new ones.

Peer-to-peer (P2P) Architecture

Peer-to-peer (P2P) architecture shares tasks and workloads among equal peers in the network, with each node acting as both a supplier and a consumer. There is no central server; each peer communicates directly with others, allowing for data and resource sharing without a central coordinator.

Pros: Highly scalable and reliable, with no single point of failure and efficient resource use.
Cons: Difficult to manage and secure, can have uneven performance, and risks losing data.

Data Flow in Peer-to-peer (P2P) Architecture

A peer initiates a request for data or a resource.
The request is broadcast to other peers in the network.
Peers with the requested data or resource respond directly.
The initiating peer receives data from multiple sources.

Data Mesh Architecture

Data Mesh architecture views data as a product managed by teams specific to each domain. Each team handles its own data and makes it accessible through standardized interfaces, while central governance ensures consistency across all data products.

Pros: Improves data quality, is easy to scale, and works well with business areas.
Cons: Requires a cultural shift, can be challenging to implement, and may lead to data duplication.

Data Flow in Data Mesh

Domain teams handle their own data products.
Data is accessed through standard interfaces.
Other teams use these products as needed.
Central management keeps everything consistent.

Data Fabric Architecture

Data Fabric architecture is a complete architecture that connects data across various environments. It uses smart, automated systems to access data where it is or move it as needed. This ensures consistent capabilities across local, cloud, and edge devices for analytics, data science, and management.

Pros: Simple integration, works well in hybrid environments, and provides automated management.
Cons: Complicated to set up, requires a significant investment, and needs specialized skills.

Data Flow in Data Fabric

Data sources connect to the fabric.
The fabric finds and organizes the data.
Users ask for the data they need.
The fabric manages access to the data.
The requested data is sent to the user.

Save or share this article