Indirect VXLAN Bridging for Scalable Data Center Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional network elements, such as routers and multilayer switches, face challenges in efficiently managing and forwarding packets to a vast number of virtual tunnel end points (VTEPs) due to the need for extensive storage of remote VTEP IP addresses, which becomes impractical as data centers scale.
Innovation Solution
The implementation of indirect VXLAN bridging, which allows network elements to reduce the number of required VTEP IP addresses by using the VXLAN protocol to encapsulate packets with intermediate routing destination addresses, enabling efficient packet forwarding across network elements without needing to store information about all VTEPs on hosts, thereby allowing packets to reach any arbitrary destination VTEP.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If network elements store remote VTEP IP addresses for direct routing, then packet delivery to VTEPs is achieved, but the storage requirement becomes impractical as data centers scale to vast numbers of egresses
Solution Approach 1:
The patent introduces an intermediary VXLAN tunnel mechanism where packets are first encapsulated and forwarded to an intermediate VTEP address, then decapsulated and re-encapsulated for final delivery. This intermediary approach allows network elements to route packets without storing comprehensive VTEP address information, resolving the contradiction between delivery capability and storage requirements
Solution Approach 2:
The patent segments the direct routing path into multiple hops through intermediate VTEPs. Instead of requiring a single direct route from source to destination VTEP, the path is divided into manageable segments that can be handled by different network elements, each with limited address storage requirements
2Adaptability or versatility
If network elements store comprehensive VTEP address information, then complete routing capability is maintained, but device complexity and scalability are compromised
Solution Approach 1:
By introducing intermediate VTEPs as mediators in the routing path, the patent enables network elements to achieve complete routing capability without maintaining comprehensive address information locally. Each network element only needs to know how to reach intermediate VTEPs, not all final destination VTEPs
Solution Approach 2:
The patent adds an additional dimension to the routing architecture by introducing an intermediate layer between source and destination VTEPs. This dimensional change transforms the routing problem from requiring global address knowledge to requiring only local knowledge of intermediate hop addresses
Data Source
AI summary
A method for transmitting MAC frames between hosts/remote machines and virtual machines across network elements (e.g., switches, routers, and multilayer switches) that conventionally do not hold capacity to address VXLAN encapsulation to any and all possible destination VTEPs within expanding data centers. More specifically, the method permits a network element the functionality of retaining VXLAN encapsulation table entries corresponding to VTEPs on Top of Rack (ToR) switches versus to VTEPs on hosts that reside under those ToR switches. This use of indirect VXLAN bridging may reduce the number of required VTEPs stored on a network element for the purposes of performing VXLAN encapsulation, thereby once again establishing the capability for packets to reach any arbitrary destination VTEP as data centers scale.