Virtual Machine Resource Allocation via Iterative Benchmark Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current resource management systems in cloud services fail to construct Virtual Machines (VMs) that meet user performance requirements, as they only consider individual component performances without accounting for cooperative performance among components like CPU, memory, and storage.
Innovation Solution
A resource management apparatus that calculates a benchmark score based on user usage and requested performance, predicts necessary component configurations, and iteratively adjusts resource allocation to ensure the VM meets the user's performance requirements by activating the VM with benchmark software and comparing measurement results to the required score.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a VM is constructed based only on individual component specifications (CPU cores, memory amount, storage amount), then the resource allocation process is simple, but the VM performance does not meet user requirements because cooperative performance among components is not considered
Solution Approach 1:
The system measures the actual benchmark score of the constructed VM and compares it with the user-required score. Based on this feedback, the system iteratively adjusts the resource configuration (CPU cores, memory, storage) to achieve the target performance level. This closed-loop feedback mechanism ensures that the final VM configuration meets the user's performance requirements.
Solution Approach 2:
The system performs a preliminary construction of the VM with initial resource allocations before actual use. This preliminary VM is then benchmarked and evaluated, allowing the system to identify performance deficiencies and make necessary adjustments before the VM is deployed for production workloads.
2Measurement precision
If the system iteratively adjusts resource configuration to meet performance requirements, then the VM performance accuracy is improved, but the construction time and complexity increase
Solution Approach 1:
The system performs iterative adjustments only when the current benchmark score does not meet the user requirements. If the initial configuration is sufficient, no further iterations are performed. This partial action approach avoids unnecessary iterative cycles and reduces the time loss while still achieving the required performance precision when needed.
Solution Approach 2:
The resource configuration is made dynamic and adjustable during the construction process. The system can modify CPU cores, memory amount, and storage amount based on benchmark results, allowing the configuration to evolve from an initial estimate to an optimized final state that meets performance targets.
3Device complexity
If the system considers only individual component performances, then the resource allocation is straightforward, but the overall VM performance is insufficient because cooperative performance among components is not evaluated
Solution Approach 1:
The system merges the evaluation of individual component performances with the evaluation of their cooperative performance. By running benchmark tests on the complete VM configuration rather than on individual components in isolation, the system captures the synergistic effects and interactions between CPU, memory, storage, and other components, leading to more accurate performance assessment.
Data Source
AI summary
A resource management apparatus includes a memory and a processor coupled to the memory. The processor is configured to receive a usage of a target virtual machine provided to a user and a request benchmark score requested by the user. The processor is configured to determine first resources to be used for constructing the target virtual machine based on the received usage and request benchmark score. The processor is configured to construct a first virtual machine using the first resources.