Cluster Resource Allocation Weight Scores
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional resource allocation techniques in computer clusters do not automatically allow clients to inherit resource allocation priority from their parent resource pools, leading to sibling rivalry and resource deprivation issues, where clients with higher resource requirements can deprive other clients of resources, causing denial of service.
Innovation Solution
A method and system that generate resource allocation weight scores for resource nodes in a cluster resource allocation hierarchy based on the number of powered-on clients, ensuring that clients inherit the resource allocation priority of their parent resource pools and guarantee a minimum amount of resources, thereby preventing resource deprivation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional resource allocation techniques are used, then resource allocation is simple, but clients cannot inherit parent resource pool priority and suffer from sibling rivalry
Solution Approach 1:
The system pre-calculates and stores resource allocation weight scores for each resource node in the hierarchy before allocation occurs. These weight scores are computed based on the number of powered-on clients and their resource requirements, allowing the allocation mechanism to simply reference pre-computed values during runtime, thus achieving priority inheritance without complex real-time calculations
Solution Approach 2:
The patent implements a hierarchical resource allocation structure where parent resource pools contain child resource pools, which in turn contain clients. Each level inherits and propagates resource allocation priorities down the hierarchy through weight scores, allowing nested structures to maintain and enforce priority relationships across multiple levels of the allocation hierarchy
2Productivity
If clients with large resource requirements are placed in the same resource pool, then resource utilization increases, but other clients suffer from resource deprivation and denial of service
Solution Approach 1:
The system assigns different resource allocation weight scores to different resource nodes based on their specific needs and the number of powered-on clients they contain. This allows each node to receive appropriate resource allocation relative to its local requirements, preventing any single node from monopolizing resources while maintaining high overall utilization
Solution Approach 2:
The system continuously monitors the number of powered-on clients in each resource node and dynamically adjusts resource allocation weight scores based on this feedback. When a node has many powered-on clients, its weight score increases, ensuring it receives sufficient resources, while nodes with fewer clients receive proportionally less, preventing resource deprivation
Data Source
AI summary
System and method for performing resource allocation for a host computer cluster use resource allocation weight scores for resource nodes in a cluster resource allocation hierarchy of the host computer cluster based on the number of powered-on clients in the resource nodes.