DF Election via Active Link Count in EVPN PE Failure
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for electing a designated forwarder (DF) in Ethernet virtual private networks (EVPNs) are not instantaneous, leading to delays in forwarding broadcast, unknown unicast, and multicast traffic, and can result in 'black-holing' of traffic and split-brain scenarios due to the reliance on timers and lack of consideration for link capacity and status.
Innovation Solution
The method determines the DF based on the number of active links between the customer edge (CE) and provider edge (PE) devices, using the link aggregation control protocol (LACP) to ensure that the PE with the most active links is elected as the DF, thereby minimizing delays and preventing duplicate traffic forwarding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional timer-based DF election methods are used, then the election process is simple to implement, but the election speed is slow causing traffic delays and black-holing
Solution Approach 1:
The patent applies preliminary action by pre-establishing the DF election criteria based on link capacity and status before actual traffic forwarding is needed. PEs continuously monitor and evaluate their link characteristics to the CE, maintaining updated information about their suitability as DF. This pre-computation eliminates the need for slow timer-based elections when traffic arrives, as the DF can be immediately determined based on pre-assessed link quality metrics.
Solution Approach 2:
The patent replaces the mechanical timer-based election system with a link-capacity-driven selection mechanism. Instead of using time-based waiting periods and sequential timer expirations, the system directly evaluates link characteristics (bandwidth, status, utilization) to determine DF eligibility. This substitution of the election mechanism with a capacity-based metric system enables instantaneous DF determination without temporal delays.
2Reliability
If timer-based DF election is used, then the implementation is straightforward, but it causes split-brain scenarios and duplicate traffic forwarding
Solution Approach 1:
The patent changes the fundamental parameter used for DF election from time-based metrics to link capacity and status metrics. By evaluating objective parameters such as link bandwidth, link status (up/down), and link utilization, the system determines DF eligibility based on physical link characteristics rather than temporal sequences. This parameter transformation ensures that only the PE with the optimal link capacity to the CE can become DF, preventing split-brain scenarios where multiple PEs might simultaneously believe they are the DF.
Solution Approach 2:
The patent implements feedback mechanisms where PEs continuously monitor their link status and capacity to the CE, and this information feeds into the DF election decision. The system uses real-time feedback about link quality to dynamically adjust DF eligibility, ensuring that the DF selection always reflects current network conditions. This feedback loop prevents duplicate traffic forwarding by ensuring that link capacity changes are immediately reflected in DF status.
3Productivity
If link capacity considerations are added to DF election, then traffic forwarding efficiency is improved, but the election process becomes more complex
Solution Approach 1:
The patent applies universality by using the same link capacity evaluation mechanism for multiple purposes: DF election, traffic load balancing decisions, and failover determination. The link capacity metrics serve as a universal criterion across different network operations, eliminating the need for separate complexity-heavy mechanisms for each function. This multi-functional use of link capacity information improves productivity while managing device complexity through reuse of the same evaluation framework.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method is implemented by a network device functioning as a provider edge (PE) device to determine a designated forwarder (DF) for sending broadcast, unknown unicast or multicast (BUM) traffic to a customer edge (CE) device, which is connected to a plurality of PE devices in an Ethernet virtual private network (EVPN) instance on a given Ethernet segment. The method reduces a time for DF determination and loss of traffic for the CE device. The method determines whether a number of active links from the network device to the CE device is greater than an active link minimum. The method also determines that the network device is the DF in response to determining that the number of active links is greater than the active link minimum.