Redundant Network Shared Switch for Cluster Bandwidth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional computer cluster networks with fat-tree topology require duplicate networks or dual-rail configurations to achieve higher bandwidth and redundancy, which significantly increase costs due to the need for twice the number of switches.
Innovation Solution
A network cluster configuration that reduces the number of switches by using two sets of switches with each client node connected to at least two separate communication paths through the switches, allowing for redundant communication paths and increased switch functionality, thereby achieving higher bandwidth and redundancy at a lower cost.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If duplicate fat-tree networks or dual-rail configurations are used to achieve higher bandwidth and redundancy, then reliability and bandwidth are improved, but cost and device complexity increase significantly due to requiring twice the number of switches
Solution Approach 1:
The patent merges two separate fat-tree networks into a single unified network fabric where switches from the first set and second set interconnect to form redundant communication paths. Instead of maintaining two independent duplicate networks, the invention combines them into one integrated structure where redundancy is achieved through multiple paths within the same network fabric, reducing the total number of switches required while maintaining reliability.
Solution Approach 2:
Switches in the unified network fabric perform multiple functions: they serve as connection points for client nodes, provide redundant communication paths, and enable both single-rail and dual-rail operational modes. Each switch can function in different capacity depending on the operational mode, eliminating the need for dedicated redundant switches and reducing overall device complexity.
2Productivity
If duplicate fat-tree networks or dual-rail configurations are used to achieve higher bandwidth, then bandwidth is improved, but cost and device complexity increase significantly
Solution Approach 1:
The patent merges two separate fat-tree networks into a single unified network fabric where switches from the first set and second set interconnect to form redundant communication paths. Instead of maintaining two independent duplicate networks, the invention combines them into one integrated structure where redundancy is achieved through multiple paths within the same network fabric, reducing the total number of switches required while maintaining reliability.
Solution Approach 2:
The network fabric dynamically adapts its operational mode between single-rail and dual-rail configurations. The system can switch between using one set of switches or both sets of switches depending on the required bandwidth and redundancy level, allowing flexible resource utilization without requiring permanent duplication of all network components.
3Reliability
If dual-rail network configurations are used to achieve higher availability, then reliability is improved, but cost increases to double that of single-rail networks
Solution Approach 1:
Switches in the unified network fabric perform multiple functions: they serve as connection points for client nodes, provide redundant communication paths, and enable both single-rail and dual-rail operational modes. Each switch can function in different capacity depending on the operational mode, eliminating the need for dedicated redundant switches and reducing overall device complexity.
Solution Approach 2:
The network fabric dynamically adapts its operational mode between single-rail and dual-rail configurations. The system can switch between using one set of switches or both sets of switches depending on the required bandwidth and redundancy level, allowing flexible resource utilization without requiring permanent duplication of all network components.
Data Source
Figure 1~4B
Figure 2
Figure 3
AI summary
In one embodiment, a computer cluster network includes at least three switches communicatively coupled to respective at least one client nodes. At least one of the at least three switches communicatively couples together at least two other ones of the plurality of switches. In a method embodiment, a method of networking client nodes includes communicatively coupling each switch of at least three switches to respective at least one client nodes. The method also includes communicatively coupling together at least two switches of the at least three switches through at least one other switch of the at least three switches.