Cascading Resource Provisioning for Data Center QoS
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face challenges in efficiently provisioning resources for multiple jobs in a shared environment while meeting quality-of-service (QoS) requirements and optimizing energy usage, as existing systems struggle to effectively consolidate resources and manage resource allocation over time intervals.
Innovation Solution
A system that establishes resource-usage models for jobs, ranks them based on QoS requirements, and provisions resources by distributing unused reservations to higher-ranked jobs, using a cascading algorithm to ensure resource needs are met across multiple time intervals, allowing for both efficient resource sharing and isolation control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If resources are allocated to meet QoS requirements for all jobs, then job quality of service is improved, but resource utilization efficiency deteriorates
Solution Approach 1:
The system performs preliminary resource allocation by establishing resource-usage models and making reservations for jobs in advance, sorted by QoS rankings. Higher-ranked jobs receive reservations first, ensuring their QoS requirements are met before allocating to lower-ranked jobs, thus efficiently managing limited resources while satisfying critical QoS demands.
Solution Approach 2:
The patent applies local quality by differentiating resource allocation based on individual job QoS requirements. Each job receives a tailored number of shares proportional to its QoS ranking, allowing high-priority jobs to receive more resources while low-priority jobs receive fewer, optimizing overall resource utilization while meeting diverse QoS needs.
2Productivity
If multiple jobs are consolidated on shared physical resources, then data center capacity is improved, but resource allocation complexity increases
Solution Approach 1:
The system segments the resource allocation process into distinct phases: establishing resource-usage models, ranking jobs by QoS requirements, making initial reservations sorted by rank, and then redistributing unused reservations. This segmentation simplifies the complex task of consolidating multiple jobs on shared physical resources by breaking it down into manageable steps.
Solution Approach 2:
The patent introduces an intermediary redistribution mechanism that mediates between initial resource allocation and final resource distribution. Unused reservations from higher-ranked jobs are redistributed to lower-ranked jobs based on their share counts, acting as a mediator to balance resource allocation while maintaining consolidation benefits.
3Reliability
If reservations are made for higher QoS ranked jobs first, then QoS satisfaction is improved, but available resources for other jobs deteriorates
Solution Approach 1:
The system discards unused reservations from higher-QoS jobs and recovers them for redistribution to lower-QoS jobs. By establishing initial reservations for higher-ranked jobs and then redistributing any unused portions to lower-ranked jobs based on their share counts, the system ensures QoS satisfaction for critical jobs while maximizing resource utilization for all jobs.
Data Source
AI summary
One embodiment of the present invention provides a system for provisioning physical resources shared by a plurality of jobs. During operation, the system establishes resource-usage models for the jobs, ranks the jobs based on quality of service (QoS) requirements associated with the jobs, and provisions the jobs for a predetermined time interval in such a way that any unused reservations associated with a first subset of jobs having higher QoS rankings are distributed to other remaining jobs with preference given to a second subset of jobs having a highest QoS ranking among the other remaining jobs. Provisioning the jobs involves making reservations for the jobs based on the resource-usage model and corresponding QoS requirements associated with the jobs.