Server Switch with Virtual Ports for VM Packet Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multiple core server systems, increasing the number of virtual machines can lead to packet throughput bottlenecks and high CPU utilization due to inadequate network bandwidth allocation and resource management among virtual machines.
Innovation Solution
A switch architecture is integrated into the server platform, featuring a link scheduler for network bandwidth allocation, VPorts for direct access to virtual machines, and a VPort swapper to manage memory and packet bursts, ensuring efficient packet switching and quality of service across virtual machines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the number of virtual machines is increased to improve resource utilization, then resource utilization improves, but packet throughput bottleneck and CPU utilization increase
Solution Approach 1:
A server switch is introduced as an intermediary device between virtual machines and the network. The switch handles packet routing and switching operations, freeing the CPU from packet processing tasks. This mediator approach resolves the contradiction by enabling multiple VMs to communicate efficiently without creating CPU bottlenecks, thus improving resource utilization while maintaining packet throughput.
Solution Approach 2:
The network processing function is segmented from the CPU and assigned to a dedicated server switch. This segmentation allows the CPU to focus on computation while the switch handles network packet operations. By dividing the system into separate functional components, the patent enables increased virtual machine density without proportionally increasing CPU utilization, thereby resolving the throughput bottleneck issue.
2Productivity
If the number of virtual machines is increased to improve resource utilization, then resource utilization improves, but CPU utilization increases
Solution Approach 1:
The server switch acts as a mediator that offloads network processing tasks from the CPU. By handling packet routing, switching, and queue management centrally, the switch enables multiple VMs to share network resources without proportionally increasing CPU workload. This resolves the contradiction by decoupling resource utilization from CPU utilization.
Solution Approach 2:
The server switch provides self-service network management capabilities including automatic packet routing, bandwidth allocation, and quality of service enforcement. These self-service functions eliminate the need for CPU intervention in network packet processing, allowing the system to scale VM density without increasing CPU utilization proportionally.
3Productivity
If network bandwidth is allocated to multiple virtual machines, then resource utilization improves, but packet switching complexity increases
Solution Approach 1:
The server switch is designed as a universal device that handles multiple functions including packet routing, bandwidth allocation, quality of service management, and virtual machine networking. By consolidating these functions into a single multi-functional component, the patent reduces overall system complexity while enabling efficient resource allocation across multiple VMs.
Solution Approach 2:
Multiple network management functions are merged into a single server switch component. Instead of having separate mechanisms for routing, bandwidth control, and VM networking, these functions are combined in one integrated device. This merging approach simplifies the packet switching architecture while maintaining the ability to serve multiple virtual machines efficiently.
Data Source
AI summary
A switch, a system and operational method for packet switching between virtual machines running in a server and a network. The server comprises a switch with swappable, virtual ports. The switch routes packets to and from the various virtual machines resident in the server memory.