Port Channel Load Balancing via Quantized Port Metrics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional load-balancing schemes for port channels are agnostic to network port load, leading to uneven utilization and potential overburdening of ports when multiple elephant traffic flows hash to the same value, resulting in inefficient traffic distribution.
Innovation Solution
Assigning quantized values based on current load to each network port and using these values to weight the selection of egress ports, ensuring that ports with lower loads are more likely to be chosen for new traffic flows, thereby distributing traffic more evenly across the port channel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional hash-based load-balancing schemes are used, then traffic flow distribution is simple and deterministic, but network ports may become unevenly utilized when multiple elephant traffic flows hash to the same value
Solution Approach 1:
The patent implements dynamic load balancing by continuously monitoring network port load metrics and adjusting the selection of egress ports based on current load conditions. The system maintains flow table entries that associate traffic flows with egress ports, and when selecting egress ports for new flows, it dynamically chooses from ports with lower current load rather than using static hash-based allocation. This dynamic adaptation allows the system to respond to changing traffic patterns and prevent any single port from becoming overloaded with multiple elephant flows.
2Speed
If load-balancing selection is agnostic to port load, then the load-balancing logic is simple and fast, but ports with lower utilization are not preferentially selected
Solution Approach 1:
The system performs preliminary monitoring and measurement of network port load metrics before making load-balancing decisions. Flow table entries are pre-established with associations between traffic flows and egress ports based on initial load conditions. When new flows need egress port selection, the system quickly consults these pre-established associations and current load metrics to make informed decisions, rather than performing complex real-time analysis. This preliminary preparation maintains decision speed while enabling load-aware port selection.
Data Source
AI summary
Methods and apparatus for load balancing across member ports for traffic egressing out of a port channel are provided herein. An example method according to one implementation may include: assigning a quantized value based on current load to each of the network ports in the port channel; receiving a data packet addressed to egress through the port channel; identifying a traffic flow with which the received data packet is associated; determining whether the identified traffic flow is a new traffic flow; and selecting one of the network ports in the port channel as an egress port. Selection of the egress port may be weighted according to the quantized value of each of the network ports in the port channel.