Flow Label Assignment for Shortest Path Bridging Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Ethernet network architectures, particularly those supporting IEEE 802.1aq standard for shortest path bridging, lack effective load balancing techniques due to incompatibility with per-hop multi-path forwarding approaches, which are necessary for efficient bandwidth utilization and performance improvement.
Innovation Solution
A flow classification process is implemented at the edge of the shortest path bridging network to determine a flow label for client frames, using entropy sources like IP headers, destination-MAC addresses, and source-MAC addresses to select among multiple equal-cost paths, and a backbone virtual LAN identifier (B-VID) is attached to frames for forwarding, allowing nodes to distribute traffic evenly across available paths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If per-hop multi-path forwarding approach is used, then load balancing and bandwidth utilization are improved, but compatibility with Ethernet network architectures and IEEE 802.1aq standard is lost
Solution Approach 1:
The patent segments the load balancing function into two parts: (1) flow label assignment at the ingress node based on client frame characteristics, and (2) flow label-based forwarding at intermediate nodes. This segmentation allows Ethernet networks to achieve multi-path forwarding without requiring changes to the fundamental Ethernet architecture or IEEE 802.1aq standard, thus resolving the contradiction between improved bandwidth utilization and architectural compatibility.
Solution Approach 2:
The patent introduces flow labels as an intermediary mechanism that bridges the gap between Ethernet's traditional single-path forwarding and the desired multi-path capability. The flow label acts as a mediator that carries path selection information through the network without requiring Ethernet nodes to implement complex multi-path routing logic, thereby maintaining compatibility while enabling load balancing.
2Productivity
If flow labels are assigned at the edge of the network, then load balancing across multiple paths is achieved, but complexity in flow classification and label assignment increases
Solution Approach 1:
The patent applies local quality by making the flow classification process localized to the ingress node only, rather than requiring all network nodes to perform complex classification. The ingress node uses local information (client frame characteristics) to assign flow labels, while intermediate nodes simply forward based on these labels. This reduces overall system complexity while maintaining load balancing efficiency.
Solution Approach 2:
The flow label is assigned in advance at the ingress node before the frame enters the core network. This preliminary action simplifies the forwarding process at intermediate nodes, as they only need to perform simple label-based lookup and forwarding without engaging in complex flow classification, thereby reducing device complexity while achieving effective load balancing.
3Productivity
If entropy sources like IP headers are used for flow label calculation, then traffic distribution is randomized and improved, but dependency on specific protocol headers increases
Solution Approach 1:
The patent implements a dynamic flow label assignment mechanism that adapts to the available information in client frames. The system can use IP headers when present for entropy-based randomization, but can also fall back to other frame characteristics when IP headers are unavailable. This dynamic approach improves traffic distribution while maintaining protocol independence and versatility.
Solution Approach 2:
The patent changes the parameter used for flow label calculation based on the available data in the client frame. When IP headers are available, they are used as the entropy source; otherwise, other frame parameters such as MAC addresses are used. This parameter flexibility ensures both improved traffic distribution through randomization and adaptability to different protocol environments.
Data Source
AI summary
A flow classification process is used at the edge of the shortest path bridging network to determine a flow label for attachment to a client frame entering the network. Any of several flow labels can be assigned to a client frame traversing the network to a particular egress node, and the flow labels are used by forwarding nodes to select among multiple equal-cost paths. In several embodiments, the flow label is calculated as a function of the client frame contents, which provide an entropy source for randomizing the selection of the flow label. This entropy source comprises the Internet Protocol (IP) header in the client frame, in some embodiments, but may comprise other client frame content in other cases.