Cloud Host Evacuation via Priority Queues and Dynamic Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for handling host failures on cloud platforms are inadequate for large-scale failures, leading to chaotic evacuation and potential collapse of the entire platform, as they fail to prioritize high-value user services and manage computing resources effectively.
Innovation Solution
A method that divides hosts into groups, sets initial priority values and evacuation thresholds, and manages failed hosts and cloud hosts based on priority, dynamically adjusting priorities and disabling queues with poor evacuation capability to ensure orderly recovery and resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If mechanical evacuation is performed without restriction during large-scale host failures, then all cloud hosts can be evacuated, but evacuation disorder occurs and the entire cloud platform may collapse
Solution Approach 1:
The patent segments the cluster into multiple host groups, each with its own failed host queue. This segmentation prevents uncontrolled global evacuation by localizing failure handling to specific groups, thereby maintaining platform stability while enabling systematic evacuation control.
Solution Approach 2:
The patent introduces priority values and evacuation count thresholds as parameters to control the evacuation process. By dynamically adjusting these parameters based on cloud host importance and failure circumstances, the system achieves controlled evacuation that prevents disorder while ensuring critical services are restored first.
2Reliability
If all cloud hosts are evacuated during large-scale failure, then comprehensive recovery is achieved, but resource allocation becomes insufficient and high-value services cannot be guaranteed
Solution Approach 1:
The patent assigns different priority values to different cloud hosts based on their user value and service importance. This local differentiation enables selective evacuation and recovery prioritization, ensuring that limited computing resources are allocated to high-value services first rather than treating all hosts uniformly.
Solution Approach 2:
The patent implements evacuation count thresholds that limit the number of evacuation attempts for each cloud host. This partial action approach prevents resource exhaustion by stopping repeated evacuation attempts for hosts that cannot be recovered, thereby preserving resources for more critical recovery operations.
3Reliability
If priority-based evacuation is implemented, then high-value services are recovered first, but the system complexity increases
Solution Approach 1:
The patent pre-assigns priority values and evacuation count thresholds to cloud hosts before failures occur. This preliminary configuration simplifies the actual evacuation process by eliminating the need for complex real-time priority calculations during failures, reducing system complexity while maintaining priority-based recovery.
Data Source
AI summary
A method for handling large-scale host failures on a cloud platform includes: configuring a corresponding failed host queue for each host group; setting initial priority values and evacuation count thresholds for cloud hosts; arranging failed hosts in failed host queues according to priorities of the failed hosts, and arranging the cloud hosts in the failed hosts according to priority values of the cloud hosts; handling, by the host group, the failed hosts the cloud hosts and in the corresponding failed host queue according to an arrangement order; when evacuation of a cloud host fails, re-determining a priority value of the cloud host; detecting evacuation capability of the host group corresponding to each failed host queue, and disabling a failed host queue having poor evacuation capability; arranging failed hosts in the disabled failed host queue into remaining failed host queues; and enabling a failed host queue meeting a condition.