Ingress Traffic Splitting for Selective Randomized Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Randomized Load Balancing (RLB) and Selective Randomized Load Balancing (SRLB) face issues such as packet missequencing, buffer overflow, and inefficient handling of best-effort traffic, as they treat all traffic classes equally and do not account for traffic destinations, leading to operational expenses and network inefficiencies.
Innovation Solution
Implementing intelligent traffic splitting methods at ingress nodes by identifying packet headers, prioritizing traffic, and filling direct circuits with traffic destined for specific nodes, allowing for direct routing of best-effort traffic and reducing buffer overflow by sorting traffic based on final destinations, while minimizing missequencing through route selection based on quality-of-service constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traffic is randomly distributed among intermediate nodes in SRLB, then load balancing is achieved, but packet missequencing occurs and buffer overflow happens
Solution Approach 1:
The patent segments traffic into different classes (guaranteed traffic and best-effort traffic) and applies different routing strategies to each segment. Guaranteed traffic is routed through dedicated circuits while best-effort traffic uses random load balancing, thereby preventing packet missequencing and buffer overflow for critical traffic while maintaining load balancing benefits for non-critical traffic.
Solution Approach 2:
The patent applies different quality characteristics to different traffic segments by identifying packet headers and classifying traffic locally at each ingress node. This allows guaranteed traffic to receive priority handling with deterministic routing while best-effort traffic receives probabilistic routing, resolving the contradiction between load balancing and reliable delivery.
2Device complexity
If all traffic is treated equally in SRLB, then simple routing is maintained, but best-effort traffic consumes excessive network resources
Solution Approach 1:
The patent implements local quality differentiation by classifying traffic into guaranteed and best-effort categories at each ingress node based on packet header identification. This allows the network to apply different resource allocation policies to different traffic classes, preventing best-effort traffic from consuming excessive network resources while maintaining relatively simple routing logic through localized classification decisions.
Solution Approach 2:
The patent changes the routing parameter from uniform random distribution to class-based differential routing. By modifying the routing behavior parameter based on traffic class (guaranteed vs. best-effort), the system achieves efficient resource utilization without significantly increasing overall routing complexity, as the classification and differential treatment are implemented through standardized mechanisms.
3Ease of operation
If direct circuits are used for all traffic, then routing simplicity is maintained, but network capacity is insufficient for high variability traffic distributions
Solution Approach 1:
The patent segments the network routing into two paths: direct dedicated circuits for guaranteed traffic and random load-balanced paths for best-effort traffic. This segmentation allows the network to maintain simple direct routing for critical traffic while utilizing the capacity benefits of random load balancing for non-critical traffic, thereby increasing overall network capacity utilization without sacrificing routing simplicity for essential services.
Solution Approach 2:
The patent creates a universal routing framework that handles both guaranteed and best-effort traffic within the same SRLB architecture. The system provides multi-functionality by supporting both deterministic direct routing and probabilistic load-balanced routing through a unified mechanism, allowing the network to adapt to different traffic requirements while maintaining operational simplicity through standardized processing at ingress nodes.
Data Source
AI summary
The performance of randomized load balanced or selective, randomized load balanced networks is enhanced by using ingress traffic engineering in addition to randomized traffic splitting. By first using the capacity of all links leading to the final destination of traffic, the remaining capacity is freed up for best effort traffic. Traffic splitting rules that enhance the performance of randomized load balanced networks in terms of packet missequencing and other quality of service criteria are also described.