Dynamic Load Balancing in Data Center Multipath Routing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current load-balancing techniques in data centers, such as ECMP routing, are not entirely satisfactory in managing network traffic efficiently, leading to potential overloading and packet delays due to network failures.

Innovation Solution

An information handling system with a topology processing module and a forwarding information base module that identifies and adjusts multipath groups based on available network devices across multiple tiers, redistributing traffic to optimize load-sharing values and mitigate network failures.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If ECMP routing is used for load balancing, then network traffic can be distributed across multiple paths, but network failures can still cause overloading and packet delays

Engineering Contradiction:
Improvenetwork traffic distributionVSAvoidnetwork failure handling
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent implements a feedback mechanism where the network device continuously monitors the operational status of subsequent network devices and dynamically adjusts load-sharing values based on this feedback. When a network failure is detected, the system receives status information, recalculates load-sharing values, and redistributes traffic to healthy paths, thereby resolving the contradiction between traffic distribution and failure handling

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent transforms the static load-sharing values in traditional ECMP routing into dynamic values that automatically adjust based on real-time network conditions. The load-sharing values are continuously updated according to the operational status of network devices, enabling the system to adapt to failures and maintain reliable traffic distribution

Inventive Principle:
Principle #15Dynamics

2Ease of manufacture

If static load-sharing values are used in ECMP routing, then configuration is simple, but the system cannot adapt to network topology changes or failures

Engineering Contradiction:
Improveconfiguration simplicityVSAvoidnetwork topology adaptation
Core Design Contradiction:
Ease of manufactureVSAdaptability or versatility

Solution Approach 1:

The patent enables the network device to automatically monitor its own network environment and self-adjust load-sharing values based on detected topology changes or failures. The system receives status information about subsequent network devices, automatically recalculates optimal load distribution, and updates its forwarding behavior without manual intervention, thus achieving both simplicity and adaptability

Inventive Principle:
Principle #25Self-service

3Productivity

If equal-cost multi-path routing is implemented, then multiple paths are utilized for traffic, but unequal load distribution occurs when network devices have different capacities or failure states

Engineering Contradiction:
Improvepath utilizationVSAvoidload distribution efficiency
Core Design Contradiction:
ProductivityVSEase of operation

Solution Approach 1:

The patent changes the load-sharing parameter from fixed equal values to dynamic values that reflect the actual capacity and status of each network path. By adjusting these parameters based on monitored network conditions, the system achieves both high path utilization and efficient load distribution proportional to each path's capabilities

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS10200286B2Systems and methods for load balancing in a data center
Publication Date: 2019.02.05 DELL PROD LP
  • US10200286B2 patent drawing
  • US10200286B2 patent drawing
  • US10200286B2 patent drawing

AI summary

Methods of load-balancing in a network are provided herein. One such method includes steps of receiving network traffic at a network device, the network traffic having a destination address and of performing a look-up in a multipath group table using the destination address. The multipath group table includes entries for recursive next-hops, each recursive next hop being associated with a connected next hop in the multipath group table. The method also include a step of hashing the network traffic according to a number of recursive next-hops associated with the destination address in the multipath group table. Additional methods and information handling system are also provided.