VM Cluster Optimization via Dynamic Instance Replacement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing platforms face performance degradation due to the use of VM instances with inadequate resources, leading to increased costs and compromised Quality of Service (QoS), as existing solutions rely on scaling rather than optimizing resource allocation based on actual demand and performance metrics.
Innovation Solution
A method and system for optimizing Virtual Machine (VM) clusters by continuously monitoring performance metrics, identifying underperforming VM instances, and replacing them with equivalent instances, redistributing application components to high-performing VMs, and terminating underperforming instances to maintain optimal resource utilization and QoS.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the number of VM instances is increased to overcome performance degradation, then the Quality of Service is improved, but the cost for instantiating VM instances increases unnecessarily
Solution Approach 1:
The patent creates a copy of the deficient VMinstance (the 'replacement VMinstance') and compares its performance metrics against the original deficient instance. This copying approach allows the system to identify and replace underperforming instances without immediately increasing the overall number of instances, thereby maintaining QoS while controlling costs.
Solution Approach 2:
The patent changes the performance parameter threshold by introducing a 'target performance value range' that dynamically adjusts based on the actual performance metrics of VM instances. This allows the system to identify deficient instances more accurately and replace them only when necessary, avoiding unnecessary scaling while maintaining service quality.
2Quantity of substance
If VM instances with degraded performance are used, then the cost is reduced, but the overall performance of the VM cluster is degraded
Solution Approach 1:
The patent implements a feedback mechanism where performance metrics of VM instances are continuously monitored and compared against target values. This feedback loop enables the system to identify deficient instances and trigger replacement processes, ensuring that cost savings from using fewer instances do not compromise overall cluster performance.
Solution Approach 2:
By creating a replacement VMinstance that is a copy of the deficient instance and comparing their performance metrics, the system can confidently replace underperforming instances with freshly provisioned ones that meet performance thresholds, thereby maintaining overall cluster performance while controlling costs.
3Quantity of substance
If resource allocation is based on monitoring resource utilisation, then the cost is optimized, but VM instances with degraded performance may be allocated instead of identical instances
Solution Approach 1:
The patent changes the allocation criterion from purely resource utilization metrics to a composite evaluation that includes performance metrics within a target value range. This parameter change ensures that VM instances are allocated based on both cost efficiency and performance consistency, preventing the allocation of degraded instances while optimizing resource utilization.
Solution Approach 2:
The patent replaces the mechanical approach of simply counting resource utilization with a more sophisticated evaluation mechanism that measures actual performance metrics. This substitution allows the system to distinguish between instances that appear similar on paper but differ in actual performance, ensuring consistent service quality while maintaining cost optimization.
Data Source
AI summary
A method and a system are provided for optimising Virtual Machine (VM) instances (121) of a Virtual Machine (VM) cluster (120) in a cloud computing platform (100) to avoid the use of VM instances (121) with degraded performance. The optimisation process of the VM instances (121) comprises the steps of identifying and optimising VM instances (121) with degraded performance in VM clusters (120). The performance of the deficient VM instances may be optimised by requesting for each deficient VM instance (121) a corresponding replacement VM instance (121) to be created by the IaaS (170), and accordingly maintain the best performing VM instance from each pair of deficient and replacement VM instances (121).