Multicast Forwarder Selection via Bandwidth Status in EVPN
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The hash-based designated forwarder (DF) election mechanism in EVPN scenarios can lead to overload on some Ethernet links and poor bandwidth utilization, degrading load sharing performance.
Innovation Solution
A method where a PE device determines the DF of a multicast flow based on the bandwidth occupation statuses of multiple Ethernet links within an Ethernet segment, ensuring equalization of load sharing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If hash-based DF election mechanism is used, then DF election can be performed, but load sharing performance deteriorates due to overload on some Ethernet links and poor bandwidth utilization
Solution Approach 1:
The patent changes the election criterion from hash-based to bandwidth-occupation-status-based. PE devices calculate bandwidth occupation ratios for each Ethernet link and elect the DF based on which device has the lowest ratio, thereby optimizing load distribution and preventing link overload while maintaining DF election functionality
Solution Approach 2:
The patent introduces a feedback mechanism where PE devices exchange bandwidth occupation status information through BGP updates. Each PE device monitors its own bandwidth usage and shares this information with other PE devices, enabling dynamic and informed DF election decisions that adapt to current network conditions
2Device complexity
If hash-based DF election is used, then DF selection can be made, but bandwidth utilization becomes unbalanced across Ethernet links
Solution Approach 1:
The patent changes the election parameter from hash algorithm output to actual bandwidth occupation ratio. By using the formula (allocated bandwidth)/(link bandwidth) to determine election priority, the system achieves balanced bandwidth utilization across links while maintaining relatively simple device complexity through standardized BGP protocol usage
3Device complexity
If DF election is performed without considering bandwidth status, then election process is simple, but some Ethernet links become overloaded
Solution Approach 1:
The patent implements feedback by having PE devices monitor and exchange bandwidth occupation status through BGP updates. This feedback loop enables the election process to consider actual link utilization, preventing overload on specific links while maintaining reasonable process complexity through automated status collection and comparison
Solution Approach 2:
The patent changes the election parameter from static (hash-based) to dynamic (bandwidth-status-based). By using real-time bandwidth occupation ratios as the election criterion, the system achieves reliable load balancing without significantly increasing device complexity, as the status information is already available through existing BGP mechanisms
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for determining a DF of a multicast flow, a device, and a system are disclosed. In an EVPN scenario, a CE device is connected to a plurality of PE devices in a dual-homed or multi-homed manner. A first PE device is any one of the plurality of PE devices. After determining that the CE device connected to an Ethernet link joins a multicast group of a multicast flow, the first PE device determines bandwidth occupation statuses of a plurality of Ethernet links included in an ES to which the Ethernet link belongs, and then determines, as a DF of the multicast flow based on the multicast flow bandwidth occupation statuses of the plurality of Ethernet links, a PE device corresponding to an Ethernet link that occupies lowest multicast flow bandwidth. This helps improve equalization of load sharing for multicast flow transmission on an EVPN.