Adaptive Resource Allocation Policy Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques for allocating cloud computing resources are heuristic, often leading to users overpaying for resources or experiencing delays in job completion, as they do not adapt based on performance and delay in obtaining performance information.
Innovation Solution
An adaptive resource allocation scheme that selects a policy for each job based on the performance of previous jobs, updating the policy selection mechanism by evaluating metrics of utility and delay, allowing for balanced job requirements and cost optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If spot pricing option is used to purchase platform resources, then the cost of resources is reduced, but the availability and reliability of resource execution is compromised
Solution Approach 1:
The patent applies dynamics by transitioning from static, fixed pricing policies to dynamic policy selection that adapts to changing conditions. The system dynamically selects between spot pricing and on-demand pricing policies based on real-time factors such as job characteristics, resource availability, and market conditions, allowing the pricing strategy to evolve and optimize both cost and reliability outcomes
Solution Approach 2:
The patent changes the parameter of pricing policy selection from a fixed state to a variable state. By introducing a policy selection mechanism that evaluates multiple pricing options and selects the optimal policy based on job requirements and market conditions, the system transforms the static pricing parameter into a dynamic decision variable that can be adjusted to balance cost and reliability
2Reliability
If on-demand pricing option is used to purchase platform resources, then the availability and reliability of resource execution is improved, but the cost of resources increases
Solution Approach 1:
The system dynamically adjusts the mix of on-demand and spot pricing based on job characteristics and market conditions. Rather than always using on-demand pricing for reliability, the policy selection mechanism dynamically determines the appropriate pricing strategy, using on-demand pricing only when necessary to ensure reliability while minimizing overall cost
Solution Approach 2:
The patent transforms the pricing parameter from a fixed on-demand choice to a flexible variable that can be adjusted between on-demand and spot pricing. The policy selection mechanism changes the pricing parameter based on job requirements, allowing the system to optimize the balance between reliability and cost for each specific job
3Device complexity
If heuristic techniques are used for determining payment for platform resources, then the allocation process is simple, but the optimization of cost and job completion efficiency is insufficient
Solution Approach 1:
The patent implements feedback by creating a closed-loop system where the policy selection mechanism learns from past job executions. By evaluating the outcomes of previous policy selections and using this feedback to improve future policy choices, the system optimizes both cost and job completion efficiency while maintaining a manageable level of complexity through iterative learning
Data Source
AI summary
A method for adaptively allocating resources to a plurality of jobs. The method comprises selecting a first policy from a plurality of policies for a first job in the plurality of jobs by using a policy selection mechanism, allocating at least one resource to the first job in accordance with the first policy, and in response to completion of the first job, updating the policy selection mechanism to obtain an updated policy selection mechanism by using at least one processor. Updating the policy selection mechanism comprises evaluating the performance of the first policy with respect to the first job by calculating a value of a metric of utility for the first policy based on conditions associated with execution of the first job and updating the policy selection mechanism based on the calculated value and a delay of execution of the first job.