Dynamic Resource Allocation for Server SLO and Efficiency Tradeoff
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In environments where multiple applications share server resources, ensuring service level objectives (SLO) while maintaining resource use efficiency is challenging due to competition for resources, leading to a tradeoff between SLO satisfaction and resource utilization.
Innovation Solution
A resource allocation device that collects service level indicator (SLI) information from applications, calculates moving averages and standard deviations, and dynamically allocates resources by sorting applications based on their calculated resource needs, ensuring that the allocation does not exceed server capacity and adhering to predetermined SLO values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If server resources are shared by multiple applications to improve resource use efficiency, then resource utilization increases, but service level objectives may not be satisfied due to resource competition
Solution Approach 1:
The patent implements dynamic resource allocation by continuously monitoring SLI metrics and adjusting resource allocation in real-time based on actual application performance and resource utilization patterns, rather than using static allocation. This allows the system to adapt to changing conditions and maintain SLO satisfaction while optimizing resource efficiency.
Solution Approach 2:
The system changes allocation parameters dynamically by calculating moving averages and standard deviations of SLI metrics to determine optimal resource allocation amounts. This statistical approach allows the system to adjust resource allocation based on measured performance data, balancing SLO guarantees with resource efficiency.
2Reliability
If server resources are occupied through pinning to satisfy SLOs, then service level objectives are highly likely satisfied, but resource use efficiency is lowered
Solution Approach 1:
The patent replaces static resource pinning with dynamic allocation that adjusts resource assignment based on monitored SLI metrics and actual resource utilization. This allows the system to provide SLO guarantees when needed while releasing resources for other uses when applications are performing well, thereby improving overall resource efficiency.
Solution Approach 2:
The system enables applications to effectively request and receive resources based on their actual performance needs as measured by SLI metrics. Rather than pre-allocating resources through pinning, the system allows applications to self-regulate resource usage and dynamically acquire resources when performance thresholds are approached, improving resource utilization while maintaining SLO satisfaction.
Data Source
AI summary
Resource use efficiency is improved while realizing quality guarantee of an application. A resource allocation device includes a storage unit that stores resource capacity information indicating a capacity of each of server resources, an SLI information collection unit that acquires information regarding an SLI at a predetermined time interval with regard to each of a plurality of applications, and a resource allocation determination unit that calculates an allocation resource amount of each application using a moving average and a standard deviation of the acquired information regarding the SLI during a predetermined period, and determines server resources which are allocation destinations of the applications by sorting the applications in descending order of the allocation resource amounts and sequentially adding the allocation resource amounts of the sorted applications within a range which does not exceed a capacity of each server resource in descending order of the allocation resource amounts.