Host Server Network Rack Redundancy Using Multiple ToR Switches
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data center networks are vulnerable to network traffic disruptions due to Top-of-Rack (ToR) switch failures, leading to server unavailability and requiring manual workload relocation or maintenance, which can be improved by enhancing server availability through multiple ToR switch connections and load balancing.
Innovation Solution
Implementing a host-to-ToR switch advertisement protocol that allows host server computers to advertise floating IP addresses independent of ToR switches, enabling multiple connections and load balancing across ToR switches, and sharing operational status to ensure network traffic rerouting and fault tolerance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a server is connected to a single ToR switch, then the device complexity is reduced, but the network availability deteriorates when the ToR switch fails
Solution Approach 1:
The patent segments the network connection by dividing it into multiple independent paths through different ToR switches. Instead of a single connection, the server is connected to multiple ToR switches, creating separate network paths that can independently handle traffic, thereby improving reliability without significantly increasing operational complexity
Solution Approach 2:
The patent changes the network configuration parameter from a single ToR connection to multiple ToR connections. This parameter change enables the system to transition from a single point of failure architecture to a redundant architecture, improving network availability while the complexity increase is managed through automated protocols
2Reliability
If multiple ToR switches are connected to a server, then network availability is improved, but the device complexity increases
Solution Approach 1:
The patent implements self-service through automated advertisement protocols where servers autonomously advertise their presence and capabilities to multiple ToR switches. The switches and servers automatically establish connections and configure routing without manual intervention, reducing the operational complexity despite the increased number of connections
Solution Approach 2:
The patent employs feedback mechanisms where servers advertise their status and the network dynamically adjusts routing based on this feedback. The advertisement protocol provides continuous feedback about server availability and network conditions, enabling automatic failover and load balancing without complex manual management
3Reliability
If floating IP addresses are advertised independently of ToR switches, then network resilience is improved, but the protocol complexity increases
Solution Approach 1:
The patent makes the advertisement protocol multi-functional by having servers advertise not only their presence but also their floating IP addresses and capabilities in a single unified advertisement message. This universal advertisement mechanism simultaneously achieves IP assignment, server discovery, and network resilience, reducing the need for separate protocols and managing complexity
Data Source
AI summary
A host-to-ToR switch advertisement protocol can be used at a data center network edge to receive IP advertisements from hosts and convey ToR switch availability status (e.g., in service, intentionally out of service, or unintentionally out of service) back to hosts and vice versa. The host server computers can have the ability to advertise an arbitrary number of IPv4 or IPv6 prefixes to the network through the ToR switches. A single host server computer can leverage multiple instances across multiple ports to the same switch or different switches to implement load balancing (e.g., equal cost multipath (ECMP)) and fault tolerance. The host server computers benefit from an increased availability by receiving network traffic via multiple ToR switches, a single ToR switch with multiple links, or a combination of both.