VM Placement Framework for Hyper-Converged Infrastructure Network Cost Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In hyper-converged infrastructures, network traffic between workloads within a cluster is not efficiently managed, leading to scalability and efficiency issues due to high network costs and traffic volumes, which existing resource schedulers fail to adequately address.
Innovation Solution
A network-aware cost-based VM placement framework that monitors and migrates workloads to minimize network costs by ranking and relocating high-traffic workload pairs, utilizing tools like VMware Distributed Resource Scheduler (DRS) and software-defined network environments to optimize network traffic across physical hosts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional resource schedulers distribute workloads by balancing CPU and memory usage, then computing capacity is efficiently allocated, but network traffic costs between workloads are not optimized
Solution Approach 1:
The patent extends the resource scheduling parameters beyond CPU and memory to include network cost metrics. The resource scheduler now considers multiple parameters simultaneously (CPU usage, memory usage, and network traffic cost) to make placement decisions, transforming the scheduling approach from single-parameter to multi-parameter optimization.
Solution Approach 2:
The patent adds a new dimension to resource scheduling by introducing network cost as an additional optimization criterion. Instead of solely balancing computational resources in two dimensions (CPU and memory), the system now operates in three dimensions by incorporating network traffic cost, enabling more comprehensive workload placement optimization.
2Adaptability or versatility
If workloads are distributed across cluster machines to balance resource usage, then system scalability is improved, but network traffic volume between workloads increases
Solution Approach 1:
The patent applies local quality by placing workloads that communicate frequently with each other on the same physical host or in close proximity within the network topology. This localized placement strategy reduces the need for long-distance network traffic while maintaining system scalability, as workloads are strategically positioned based on their communication patterns rather than being uniformly distributed.
3Productivity
If additional hardware is assigned to tasks requiring more resources, then task performance is improved, but network infrastructure complexity and costs increase
Solution Approach 1:
The patent merges multiple workloads that have high communication requirements into the same physical host or nearby hosts within the cluster. By combining these workloads spatially, the system reduces network traffic between them and simplifies the network infrastructure requirements, as internal host communication does not require external network switches and routing infrastructure.
Data Source
AI summary
Systems and methods for reducing network cost in a hyper-converged infrastructure are disclosed. The network cost of workload pairs can be assessed. Migration of the workloads can be considered to reduce the network cost and improve the network efficiency of the hyper-converged infrastructure.