Scope-Based Hang Avoidance in Distributed Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current hang avoidance mechanisms in large-scale computing systems often result in system-wide disruptions, failing to provide targeted and efficient hang avoidance, especially in distributed architectures, leading to performance degradation and unfair resource allocation.
Innovation Solution
Implementing a system with Fast Hang Avoidance (FHA) controllers that detect resource request thresholds and activate FHA mechanisms on a scope-based level, allowing for targeted intervention within specific scopes to prevent hangs without affecting other parts of the system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If system-wide hang avoidance mechanisms are activated across all resources and requestors, then hang avoidance coverage is improved, but system performance and resource allocation fairness deteriorate due to unnecessary disruptions
Solution Approach 1:
The patent divides the computing system into multiple scopes (system-wide, domain-wide, resource-level) and implements hang avoidance mechanisms at different granularities. Each scope has its own FHA controller that independently manages hang avoidance for its specific domain, allowing targeted intervention without affecting unrelated system parts. This segmentation enables selective activation of FHA mechanisms only where hang conditions are detected, resolving the contradiction between comprehensive coverage and performance maintenance.
Solution Approach 2:
The patent applies different hang avoidance strategies and activation thresholds to different scopes and resource types. Each FHA controller configures its own activation parameters based on local conditions, requestor types, and resource characteristics. This local quality approach ensures that hang avoidance is applied appropriately and selectively in specific locations rather than uniformly across the entire system, maintaining performance while providing necessary protection.
2Reliability
If fast hang avoidance mechanisms are activated on a wide scale, then hang prevention capability is improved, but resource allocation fairness deteriorates due to unnecessary blocking of legitimate requests
Solution Approach 1:
The patent implements preliminary detection of hang conditions by monitoring resource request patterns and detecting FHA conditions before actual hangs occur. FHA controllers continuously assess whether activation thresholds are exceeded and proactively activate or deactivate FHA mechanisms based on current system state. This preliminary action allows the system to prevent hangs only when necessary, avoiding unnecessary blocking of legitimate requests and maintaining resource allocation fairness.
Solution Approach 2:
The patent employs feedback mechanisms where FHA controllers continuously monitor resource request patterns, detect hang conditions, and adjust FHA activation status accordingly. The system feedback loop evaluates whether FHA conditions are met based on activation settings, requestor types, and current system state, then adjusts FHA mechanism activation to maintain both hang prevention and resource allocation fairness dynamically.
Data Source
AI summary
A network of hang avoidance controllers and components which provide layer or scope based hang avoidance mechanisms in a distributed computing system is described. The detection of hang avoidance conditions and activation of the hang avoidance mechanisms are implemented on various limited scopes in the computing system, which prevent unnecessary system wide interruptions to avoid potential hangs in the system.