Link Aggregation Bandwidth Reservation via Component Link Sub-TLV
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In traffic-engineered networks, the existing methods face challenges in efficiently guiding specific traffic towards underutilized paths, leading to overutilization of 'shortest' paths and decreased network performance, as they rely on pseudorandom distribution of bandwidth across logical link aggregation groups (LAGs), causing uneven loading on component links and potential packet delay or dropping.
Innovation Solution
The introduction of a new sub-TLV type in traffic engineering link state advertisements allows routers to advertise and manage bandwidth on individual component links within a LAG, enabling explicit reservation and management of bandwidth on specific links based on traffic requirements, rather than relying on pseudorandom hashing, thereby allowing for more precise traffic engineering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If pseudorandom hashing is used to distribute traffic across LAG component links, then traffic distribution is simplified, but bandwidth management precision deteriorates
Solution Approach 1:
The patent segments the LAG bandwidth management into two levels: aggregate bandwidth management at the LAG level (maintaining simplicity) and component link-level bandwidth management (enabling precision). The new sub-TLV structure allows advertisement of both aggregate LAG bandwidth and individual component link bandwidths, enabling routers to make precise bandwidth decisions while maintaining simplified operational procedures.
Solution Approach 2:
The patent applies local quality by allowing different bandwidth characteristics to be defined for different component links within the same LAG. Each component link can have its own bandwidth parameters advertised through the component links sub-TLV, enabling precise local bandwidth management while the overall LAG maintains a unified management interface.
2Measurement precision
If component link bandwidth information is advertised for each link in a LAG, then traffic engineering precision is improved, but information overhead increases
Solution Approach 1:
The patent merges component link bandwidth information into a single structured sub-TLV (component links sub-TLV) that is carried within the existing TE LSA framework. This consolidation allows multiple component link bandwidth parameters to be advertised in one unified structure rather than requiring separate advertisements for each link, thereby reducing information overhead while maintaining precision.
Solution Approach 2:
The component links sub-TLV structure is designed to be universal and multi-functional, capable of carrying bandwidth information for multiple component links simultaneously. The sub-TLV can accommodate variable numbers of component link entries and supports both aggregate and individual link bandwidth parameters, making it a versatile solution that reduces overall information overhead.
3Device complexity
If pseudorandom distribution is used across LAG links, then setup complexity is reduced, but network performance deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-advertising component link bandwidth information through the new sub-TLV structure before traffic engineering decisions are made. Routers receive and store this bandwidth information in advance, enabling them to make informed path selection decisions that optimize network performance without increasing setup complexity during actual traffic flow.
Solution Approach 2:
The patent implements feedback by using advertised component link bandwidth information to guide traffic engineering decisions. Routers continuously monitor bandwidth utilization and use the pre-advertised component link characteristics to dynamically select optimal paths, creating a feedback loop that improves network performance while maintaining simple setup procedures.
Data Source
AI summary
Traffic engineering using a label-switching protocol is enhanced for label-switched paths that traverse a logical link that is an aggregation of component links. In one embodiment, a label edge router is provided with information regarding the bandwidth capabilities and loading of the component links of a LAG. The label edge router is then allowed to set up paths that traverse a specific component link of a LAG, and reserve bandwidth on such a component link. Other traffic may continue to be distributed across the LAG membership.