Cloud Appropriateness Value for HPC Job Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In High-Performance Computing (HPC) systems, there is a trade-off relationship between mean wait time and cloud cost, where reducing wait time by increasing cloud processing leads to higher costs, and existing techniques do not improve this relationship.
Innovation Solution
An information processing apparatus that calculates a cloud appropriateness value using the cost of using the cloud for a scheduling-target job and an index indicating the quantity of all jobs waiting to be scheduled, and determines whether to offload the job to the cloud based on this value.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If the amount of processing executed by the cloud is increased to reduce mean wait time, then mean wait time is reduced, but cloud cost increases
Solution Approach 1:
The system dynamically changes the cloud usage parameter based on the calculated appropriateness value, which is derived from cost and workload index. When the appropriateness value indicates high suitability, the system offloads jobs to the cloud; otherwise, it processes them locally, thereby optimizing the trade-off between wait time and cost
Solution Approach 2:
The cloud bursting strategy is made dynamic through continuous calculation of the appropriateness value for each scheduling decision. The system adapts its cloud usage behavior in real-time based on current cost conditions and workload characteristics, rather than using a fixed strategy
Data Source
AI summary
An information processing apparatus includes: a calculation unit calculates a cloud appropriateness value using a cost when a cloud is used for a scheduling-target job and an index indicative of a quantity of all jobs that are yet to be allocated to a server and waiting to be scheduled; and a determination unit determines, in accordance with the cloud appropriateness value, whether or not the scheduling-target job is to be offloaded to the cloud.


