Dynamic Resource Allocation in Virtualized Data Centers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In enterprise data centers with silo-oriented architecture, resource utilization is low due to over-provisioning, leading to service-level violations from varying application demands, especially in consolidated environments where multi-tier applications have different resource needs.
Innovation Solution
A method for dynamically controlling resource allocation across virtual machines by using a feedback controller to adjust CPU resources based on application-level metrics and service level goals, ensuring high utilization and differentiated resource allocation across multiple application stacks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If applications are consolidated in a shared infrastructure using virtualization, then resource utilization is improved, but meeting application-level quality of service (QoS) goals becomes more difficult
Solution Approach 1:
The patent implements dynamic resource allocation by continuously monitoring application performance metrics and adjusting CPU resource allocations in real-time. The feedback controller modifies resource allocations based on actual QoS performance, enabling the system to adapt to changing application demands while maintaining both high resource utilization and QoS goal achievement across multi-tier applications
Solution Approach 2:
The patent establishes a feedback control mechanism where application performance metrics are monitored and fed back to the resource allocation controller. This feedback loop enables continuous adjustment of CPU allocations to virtual machines based on actual QoS performance, resolving the contradiction by allowing the system to learn from past performance and optimize future resource distribution
2Reliability
If dedicated servers with tailored software stacks are used for each application, then application performance requirements are met, but data centers become under-utilized
Solution Approach 1:
The patent enables a universal shared infrastructure to serve multiple applications with different resource requirements. By implementing dynamic resource allocation and feedback control, the same virtualized infrastructure can adaptively serve diverse application needs, replacing the need for dedicated servers while maintaining application performance through intelligent resource distribution
3Reliability
If resource allocations are increased to meet QoS goals in consolidated environments, then service level quality is improved, but overall resource utilization efficiency decreases
Solution Approach 1:
The patent dynamically changes resource allocation parameters based on real-time application performance metrics. The feedback controller adjusts CPU allocation percentages, virtual machine resource limits, and tier-specific resource pools according to actual QoS performance, enabling the system to allocate resources efficiently rather than statically, thus improving both service level quality and allocation efficiency
Data Source
AI summary
A method for controlling resource allocation is provided. The method includes determining a service metric associated with a first application, wherein the first application is associated with one or more virtual machines. The method further includes comparing the service metric to an application specific service level goal associated with the first application and modifying a resource allocation associated with the first application at one or more of the virtual machines.