Work Distribution via Manager-Coordinated Session Pairs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing checkpoint systems are inefficient due to large data storage requirements and lack of mechanisms to track and restore inter-computer communication states, leading to data loss and interrupted messaging processes during computer failures or outages in distributed computing environments.
Innovation Solution
A system and method for distributing work across cooperating computers using client-to-server session pairs with restart data and message queues, where a manager program coordinates client and server sessions, storing and sharing request and response messages, and restart data to ensure seamless continuation of work across different computers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional checkpoint systems are used to save program state, then program recovery is possible, but data storage requirements become excessively large and inter-computer communication states cannot be restored
Solution Approach 1:
The patent extracts only the essential restart data needed for session recovery (message queues and communication state) from the complete program checkpoint, storing only this critical information externally at the manager program rather than saving the entire program state. This reduces storage requirements while maintaining recovery capability.
Solution Approach 2:
The system segments the checkpoint problem into two parts: essential restart data (message queues and communication state) stored externally at the manager, and program code stored locally. This segmentation allows recovery of communication state without storing large amounts of program data externally.
2Reliability
If checkpoint data is saved to storage, then program state can be recovered, but inter-computer communication states are not tracked and messaging processes are interrupted
Solution Approach 1:
The system performs preliminary action by continuously maintaining message queues and communication state information in the manager program before any failure occurs. This ensures that when a session needs to be restarted, the communication state is already prepared and available, eliminating the need to re-establish messaging connections.
Solution Approach 2:
The manager program acts as an intermediary that maintains message queues and communication state separately from the client-server sessions. This intermediary preserves communication state independently, allowing sessions to be restarted without losing messaging context or interrupting communication flows.
3Reliability
If complete program checkpoints are saved, then full state recovery is possible, but system complexity and storage overhead increase significantly
Solution Approach 1:
The patent extracts only the minimal necessary data (message queues and communication state) from complete program checkpoints and stores this externally at the manager. This extraction simplifies the checkpoint system by eliminating the need to manage large program state files while maintaining session continuity capability.
Solution Approach 2:
Instead of saving complete program checkpoints, the system creates and maintains simplified copies of essential restart data (message queues and communication state) at the manager program. These lightweight copies enable recovery without the complexity of full checkpoint management.
Data Source
AI summary
A system and method are presented that improve work cooperation among computers. Communication between programs is combined with restart data to travel via a manager on a third computer. Work is represented by a hierarchical set of client to server session pairs that contain restart and message data. Reliability, security, scalability, and aggregate performance are improved with minimal impact on latency of a single piece of work.