Message Queue Restoration Using ML-Guided Critical Message Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
System outages, especially unplanned ones, require substantial time and computation to re-establish the system to a normal operational state, leading to delays in servicing requests and impacting critical functions and data structures during system restart.
Innovation Solution
A computer-implemented method using a trained machine learning model to determine a targeted number of critical messages for synchronous restoration and non-critical messages for asynchronous restoration, minimizing system restart time by rebuilding only essential messages immediately upon system recovery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all messages in the message queue are restored during system restart, then message completeness is improved, but system restart time increases
Solution Approach 1:
The patent segments messages into critical and non-critical categories based on their importance to system operation. Critical messages are restored synchronously during system restart, while non-critical messages are restored asynchronously in the background. This segmentation allows the system to restore essential messages quickly without being blocked by the need to restore all messages, thereby reducing system restart time while maintaining message completeness for critical operations.
Solution Approach 2:
The patent applies partial action by restoring only the targeted number of critical messages during synchronous restoration rather than all messages in the queue. The machine learning model determines an optimal targeted number that ensures system functionality is maintained without requiring complete message restoration. This partial restoration approach significantly reduces system restart time while still providing sufficient message data for critical operations to proceed.
2Reliability
If synchronous message restoration is performed for all messages, then message availability is improved, but computational overhead increases
Solution Approach 1:
The patent segments message restoration into two phases: synchronous restoration of critical messages and asynchronous restoration of non-critical messages. By segmenting this way, the system performs computational operations only on critical messages during the synchronous phase, significantly reducing immediate computational overhead. Non-critical messages are processed asynchronously in the background, distributing the computational load over time rather than concentrating it during system restart.
Solution Approach 2:
The patent applies partial action by using machine learning to determine a targeted number of critical messages that is less than the total number of messages in the queue. This partial restoration approach reduces the computational overhead during synchronous restoration while still ensuring sufficient message availability for critical operations. The machine learning model optimizes the targeted number to balance message availability with computational resource consumption.
3Productivity
If machine learning model is used to determine targeted number of messages, then restoration efficiency is improved, but model training complexity increases
Solution Approach 1:
The patent applies preliminary action by training the machine learning model during system development and testing phases, before actual system outages occur. The model is trained on historical data about message patterns, system performance, and outage scenarios to learn optimal restoration strategies. Once trained, the model can quickly determine the targeted number of messages to restore during actual outages without requiring complex real-time training, thereby achieving high restoration efficiency while managing model training complexity through advance preparation.
Data Source
AI summary
Embodiments determine evaluation data from at least one message queue, determine a targeted number of messages included in the at least one message queue and a confidence value by using a trained machine learning model with the evaluation data, determine that the confidence value is greater than a predetermined threshold, perform synchronous message restoration based on the targeted number of messages and the confidence value being greater than the predetermined threshold, and perform remaining system restart functions in response to the synchronous message restoration being completed.