MQTT Broker Cell Matrix Architecture for Scalable Resilience

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing MQTT implementations face limitations in network performance, resiliency, and scalability due to single-node broker designs, which constrain application resiliency and scale, and introduce bottlenecks and single points of failure.

Innovation Solution

The implementation of an MQTT broker cell comprising multiple back-end and front-end brokers arranged in a matrix structure, with vertical and horizontal chains for state update replication, enabling scalable and resilient network architecture.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a single-node broker design is used, then the system is simple to implement, but network performance, resiliency, and scalability are limited

Engineering Contradiction:
Improvenetwork resiliencyVSAvoidbroker architecture complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The broker is segmented into multiple independent nodes organized in a matrix structure with m vertical chains and k horizontal chains. Each node maintains a portion of the system state, distributing the broker functionality across multiple components rather than relying on a single node. This segmentation enables the system to tolerate node failures while maintaining operational capacity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements a nested replication structure where state updates propagate through vertical chains from head to tail nodes, and horizontal chains replicate across all nodes. This nested chain structure allows each node to be part of multiple replication paths, providing redundant state copies while maintaining organized state management within the distributed architecture.

Inventive Principle:
Principle #7Nested doll (Nesting)

2Productivity

If a single-node broker is used, then the implementation is straightforward, but the system cannot scale to large networks

Engineering Contradiction:
Improvenetwork scalabilityVSAvoidsystem architecture complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The broker system is divided into m vertical chains with k nodes each, allowing the network to scale by adjusting m and k independently. This segmentation enables horizontal scaling (adding more chains) and vertical scaling (adding more nodes per chain) without requiring complete system redesign, thus supporting large network deployments.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from a single-dimension (single-node or simple cluster) architecture to a two-dimensional matrix structure with vertical and horizontal chains. This dimensional expansion allows the system to scale along multiple independent axes, providing flexible capacity expansion and load distribution that supports large-scale networks while maintaining manageable complexity through structured organization.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

3Reliability

If replication is implemented across multiple nodes, then fault tolerance improves, but replication overhead increases

Engineering Contradiction:
Improvefault toleranceVSAvoidreplication overhead
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The nested chain structure allows state updates to be replicated efficiently through organized propagation paths. Vertical chains replicate state from head to tail nodes, while horizontal chains provide additional replication across the matrix. This nested organization reduces redundant transmissions by structuring replication along defined chains rather than requiring all-to-all communication, thus lowering replication overhead while maintaining fault tolerance.

Inventive Principle:
Principle #7Nested doll (Nesting)

Solution Approach 2:

The system uses selective copying of state updates along chain structures rather than full state synchronization between all nodes. Each node copies only the necessary state portions relevant to its position in the chains, reducing the amount of data that needs to be replicated and processed compared to full state copying approaches.

Inventive Principle:
Principle #26Copying

4Reliability

If a single-node broker is used, then the system requires no third-party storage, but the broker becomes a single point of failure

Engineering Contradiction:
Improveelimination of single point of failureVSAvoidbroker structure complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The broker functionality is segmented across multiple nodes in the matrix structure, eliminating the single point of failure inherent in single-node designs. Each node in the vertical and horizontal chains can continue operating independently if other nodes fail, providing fault tolerance without requiring external third-party storage systems.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent combines the storage and computation functions within the broker nodes themselves, merging the role of message broker with distributed state storage. This integration eliminates the need for separate third-party storage systems while distributing the storage burden across multiple broker nodes, thereby achieving both fault tolerance and architectural simplicity.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP4342163B1Broker cell for distributed message system
Publication Date: 2025.03.05 MICROSOFT TECHNOLOGY LICENSING LLC
  • EP4342163B1 patent drawingFigure 1
  • EP4342163B1 patent drawingFigure 2A
  • EP4342163B1 patent drawingFigure 2B

AI summary

Examples are disclosed that relate to message queuing telemetry transport (MQTT) broker cells. One example provides a computing system configured to implement an MQTT broker cell comprising instructions executable to operate two or more back-end brokers arranged in a matrix, the matrix comprising m vertical chains of back-end brokers and k back-end brokers in each vertical chain, each vertical chain comprising at least a head back-end broker and a tail back-end broker, each vertical chain configured to replicate a state update received at the head back-end broker through the vertical chain to the tail back-end broker, and operate n front-end brokers, each front-end broker configured to output a control message to a selected vertical chain of the m vertical chains and to output an application message for publication to subscribers and to one or more other MQTT broker cells. The instructions are further executable to operate r networking devices.