L3 Gateway for VXLAN Virtual Switches
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data center networking solutions, such as VLANs and VXLAN, face limitations in scalability and mobility due to bottlenecks and the need for reconfiguration when virtual machines (VMs) move, especially under heavy data traffic loads and in cloud computing environments.
Innovation Solution
Implementing a virtual switch as the default gateway for VMs, which assigns virtual MAC and IP addresses and swaps them with physical MAC addresses for routing, allowing packets to be forwarded through any attached routing component without requiring all traffic to pass through a single gateway, and designating a single virtual switch to respond to address requests to reduce network traffic and prevent ARP storms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all data traffic passes through a single VXLAN server or VPN gateway, then network isolation and routing control are achieved, but network bottlenecks occur under heavy traffic loads
Solution Approach 1:
The patent segments the centralized gateway function by allowing multiple virtual switches to act as default gateways for VMs. Instead of all traffic converging at a single VXLAN server or VPN gateway, each host's virtual switch can independently perform routing functions, dividing the traffic load across multiple nodes in the network.
Solution Approach 2:
The patent transitions from a single-point gateway architecture to a distributed gateway architecture where routing functionality is embedded at multiple virtual switch locations. This dimensional change from centralized to distributed architecture eliminates the single bottleneck while maintaining network isolation through virtual switching.
2Device complexity
If a single VPN gateway or VXLAN server is used for all traffic, then routing control is simplified, but VM mobility requires frequent reconfiguration
Solution Approach 1:
The patent enables virtual switches to autonomously perform address request handling and routing decisions without requiring centralized reconfiguration. When a VM moves to a new host, the local virtual switch can independently respond to address requests and update routing tables, eliminating the need for manual reconfiguration and supporting seamless VM mobility.
Solution Approach 2:
The patent implements dynamic routing where virtual switches can adapt their routing behavior based on current network conditions and VM locations. The system transitions from static, pre-configured routing to dynamic, on-demand routing that automatically adjusts when VMs move, enhancing adaptability while maintaining manageable complexity.
3Adaptability or versatility
If multiple virtual switches respond to address requests, then VM mobility is supported, but ARP storms and network traffic increase
Solution Approach 1:
The patent implements a feedback mechanism where virtual switches monitor address requests and coordinate their responses. When multiple virtual switches are capable of responding to an address request, they use feedback signals to determine which switch should actually respond, preventing duplicate responses and reducing network traffic while still supporting VM mobility.
Solution Approach 2:
The patent introduces asymmetric response behavior among virtual switches. Instead of all virtual switches responding symmetrically to address requests, the system designates specific virtual switches to respond based on their current state and network position. This asymmetric approach reduces the number of responses while maintaining mobility support.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In an embodiment, the disclosure includes an apparatus comprising a host component which comprises a virtual switch associated with a virtual machine (VM). The host component is in communication with a routing component which is connected to a network and comprises a network address. The VM sends a communication packet to the virtual switch. The communications packet comprises address information comprising a virtual destination address. The virtual switch replaces the virtual destination address of the communications packet with the network address of the routing component. The disclosure also includes a network comprising a plurality of host components each comprising a virtual switch. The host components are connected to a common routing component. A first virtual switch is designated to respond to address requests and all non-designated virtual switches are configured not to respond to address requests.