Commodity Switch Mesh for Data Center Traffic Engineering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data center architectures face challenges such as resource fragmentation, poor server-to-server connectivity, and reliance on expensive, proprietary hardware that scales up rather than out, leading to bottlenecks and increased risk of failures.
Innovation Solution
A simple mesh-like architecture using commodity switches with modified control planes for flexible traffic engineering and multipath routing, disaggregating load balancing into standard servers, allowing any server to be part of a pool and optimizing bandwidth between all servers, enabling scalable and resilient data center networking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional IP routers and load balancers are used to span traffic across VLANs, then server connectivity is provided, but expensive customized hardware creates bottlenecks and increases cost
Solution Approach 1:
The patent replaces expensive proprietary IP routers with commodity Ethernet switches that copy the routing functionality. By using standard Ethernet switching with VLAN tagging and routing protocols, the system achieves router-like connectivity without the cost and complexity of customized hardware, allowing any-rack-to-any-rack server communication through standard network infrastructure
Solution Approach 2:
The patent makes commodity Ethernet switches perform multiple functions previously requiring specialized hardware. These switches handle VLAN routing, load balancing, and server-to-server connectivity, replacing the need for separate IP routers, load balancers, and specialized networking equipment with a universal switching platform that scales economically
2Device complexity
If traffic is concentrated in a few network routers and switches, then routing control is simplified, but the system becomes vulnerable to failures and outages
Solution Approach 1:
The patent segments the network into multiple distributed Ethernet switching domains instead of relying on a few centralized routers. Each commodity switch operates independently with full VLAN routing capability, creating redundant paths throughout the network. This segmentation ensures that failure of any single switch does not cause system-wide outages, as traffic can be routed through alternative switches and paths
Solution Approach 2:
The patent changes the network architecture parameter from centralized routing control to distributed switching control. By enabling routing functionality at the Ethernet switch level rather than requiring centralized IP router control, the system achieves both simplified local control and improved reliability through redundancy. Multiple identical commodity switches can handle the same routing tasks, providing failover capability
3Device complexity
If conventional VLAN architecture is used, then network segmentation is achieved, but bandwidth constraints limit data intensive applications
Solution Approach 1:
The patent adds a new dimension to network segmentation by implementing routing functionality within the Ethernet switching plane itself, rather than relying on hierarchical VLAN boundaries enforced by external routers. This allows traffic to be segmented and routed at the Ethernet level with full bandwidth utilization, eliminating the bottleneck where VLAN-spanning traffic must pass through constrained IP router ports
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system for commoditizing data center networking is disclosed. The system includes an interconnection topology for a data center having a plurality of servers and a plurality of nodes of a network in the data center through which data packets may be routed. The system uses a routing scheme where the routing is oblivious to the traffic pattern between nodes in the network, and wherein the interconnection topology contains a plurality of paths between one or more servers. The multipath routing may be Valiant load balancing. It disaggregates the function of load balancing into a group of regular servers, with the result that load balancing server hardware can be distributed amongst racks in the data center leading to greater agility and less fragmentation. The architecture creates a huge, flexible switching domain, supporting any server/any service, full mesh agility, and unregimented server capacity at low cost.