TRILL Cloud Switch Leaf Switch Connectivity Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The implementation of the TRILL protocol with cloud switch architectures faces inefficiencies due to suboptimal inter-cloud switch domain forwarding and inefficient next-hop selection, particularly because TRILL's link state routing protocol is not aware of individual leaf switches within cloud switch domains and their interconnections, leading to increased latency and bandwidth consumption.
Innovation Solution
The solution involves adding indications of connectivity between leaf switches and RBridges to link state packets and optionally including host MAC addresses and VLAN information in ESADI frames, allowing the SPF and distribution tree computation algorithms to calculate optimal next-hop leaf switches within cloud switch domains, and collecting traffic engineering parameters for traffic-engineered multipathing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If TRILL protocol is implemented with cloud switch architecture, then forwarding capabilities and multipathing are improved, but inter-cloud switch domain forwarding becomes suboptimal due to lack of awareness of individual leaf switches
Solution Approach 1:
The patent segments the cloud switch domain information into individual leaf switch connectivity details. By adding Type-Length-Value (TLV) objects to Link State Packets that contain specific leaf switch connectivity information, the system divides the aggregated cloud switch domain view into granular leaf switch-level views, enabling optimal path calculation through individual leaf switches rather than treating the entire domain as a single entity.
Solution Approach 2:
The patent introduces an intermediary mechanism in the form of modified Link State Packets with extended TLV objects. These packets act as intermediaries that carry detailed leaf switch connectivity information between cloud switch domains, allowing the Shortest Path First algorithm to make informed forwarding decisions based on actual leaf switch interconnections rather than aggregated domain information.
2Adaptability or versatility
If TRILL's link state routing protocol is used, then routing within Layer-2 framework is achieved, but next-hop selection becomes inefficient due to lack of awareness of individual leaf switches and their interconnections
Solution Approach 1:
The patent applies preliminary action by pre-calculating and distributing detailed leaf switch connectivity information through modified Link State Packets before forwarding decisions are made. The TLV objects contain pre-computed connectivity data that enables the Shortest Path First algorithm to efficiently determine optimal next-hops without requiring real-time discovery or additional lookup operations.
3Ease of operation
If cloud switch domains are treated as single entities, then administration is simplified, but forwarding paths cannot be optimized for individual leaf switch connections
Solution Approach 1:
The patent applies local quality by enabling different levels of granularity for different purposes. At the administration level, cloud switch domains remain as single entities for simplicity. At the forwarding level, the system utilizes detailed leaf switch connectivity information from TLV objects to optimize individual paths. This allows simultaneous maintenance of administrative simplicity and forwarding optimization through context-dependent information usage.
Data Source
AI summary
In one embodiment, a plurality of leaf switches that include host facing ports are configured as a cloud switch. An indication of connectivity between the leaf switches of the cloud switch and routing bridges (RBridges) external to the cloud switch may be added to link state packets (LSPs) sent over the at least one logical shared media link. A lookup table may be generated that specifies next hop leaf switches. The generated lookup table may be used to forward frames to one or more particular nexthop leaf switches. Further, traffic engineering parameters may be collected. Equal cost multipath (ECMP) nexthop leaf switches and distribution trees to reach one or more destinations may be examined. Traffic may be distributed across ones of them based on the traffic engineering parameters.