Ephemeral Logical Message Topics for Low-Load Queue Scaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional messaging middleware systems face inefficiencies in scaling messaging resources, leading to perpetual queue rebalancing operations that consume significant computing resources, especially during diurnal load cycles, and fail to conserve resources by rebalancing message queues.
Innovation Solution
A messaging middleware platform that dynamically scales message queue resources by creating new physical topics and reconfiguring message producers and consumers to redirect message flows without transferring messages among queues, allowing existing consumers to continue processing until old topics are fully processed or expire, thus eliminating the need for queue rebalancing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If message queue rebalancing is performed during scaling operations, then message distribution is improved, but computing resources are significantly consumed
Solution Approach 1:
The system segments message routing by introducing multiple physical topics (topic-A, topic-B, topic-C) that correspond to different message queue groups. Each physical topic is independently managed with its own message brokers, allowing selective activation without affecting other segments. This segmentation enables the system to route messages to specific active segments based on load conditions, avoiding the need to rebalance across all queues.
Solution Approach 2:
The patent introduces an intermediary layer (the workflow manager and topic mapping mechanism) that sits between message producers and the physical message queues. This intermediary dynamically maps logical topics to physical topics based on current system state, enabling load distribution without direct intervention in the message queues themselves. The intermediary handles the complexity of scaling operations, isolating the message processing system from rebalancing activities.
2Productivity
If message queues are rebalanced during scaling, then load distribution is optimized, but message processing continuity is disrupted
Solution Approach 1:
The system performs preliminary actions by pre-configuring multiple physical topics and their associated message brokers before scaling is needed. When scaling occurs, the system simply activates pre-prepared infrastructure rather than creating and balancing queues in real-time. Message producers are预先 configured with multiple physical topic references, enabling seamless failover and load distribution without disrupting ongoing message processing.
Solution Approach 2:
The patent changes the routing parameter from direct queue assignment to dynamic topic mapping. Instead of fixing messages to specific queues, the system uses configurable parameters (physical topic references) that can be dynamically adjusted. The workflow manager monitors system state and changes the active physical topic parameter based on broker availability and load conditions, maintaining continuous message processing while optimizing distribution.
3Productivity
If new message brokers are activated during scaling, then system capacity is increased, but infrastructure complexity increases
Solution Approach 1:
The patent implements universality by designing physical topics and message brokers that can serve multiple functions. The same infrastructure components (message brokers, topics) are used for both normal operations and scaling scenarios. The workflow manager provides a unified control mechanism that manages both small and large scale operations through the same interface, reducing the need for specialized infrastructure for different scaling levels.
Data Source
AI summary
A messaging middleware platform implemented on a computer system. The platform comprises a directory service application that maps logical to physical topic references, where the physical topic references refer to message topics instantiated on message broker instances, a plurality of message producer instances that execute on the computer system, where each message producer instance stores messages on message topics instantiated based on a logical topic reference built into the message producer and dereferenced to a physical topic reference at run-time, and a workflow manager application that executes on the computer system that creates new message broker instances when the message depth exceeds a predefined threshold, revises associations between logical topic references and physical topic references in the directory service application, and causes the message producer instances to be reconfigured to dereference the logical topic reference to a physical topic reference associated with the new message broker instances.