- Choose an unused
VLAN (also known
Scroll Indexterm secondary switched virtual interface primary VLAN
Scroll Indexterm secondary SVI primary switched virtual interface
a switched virtual interface or SVI here).
Scroll Indexterm secondary switched virtual interface primary SVI
- Assign the SVI an unrouteable link-local address to give the peer switches layer 3 connectivity between each other.
Configure the VLAN as a VLAN subinterface on the peer link bond rather than the VLAN-aware bridge, called peerlink. If you're configuring the subinterface with NCLU, the VLAN subinterface is named 4094 by default (the subinterface named peerlink.4094 below). If you are configuring the peer link without NCLU, Cumulus Networks still recommends you use 4094 for the peer link VLAN if possible. This ensures that the VLAN is completely independent of the bridge and spanning tree forwarding decisions.
- Include untagged traffic on the peer link, as this avoids issues with STP.
- Optionally, you can specify Specify a backup interface, which is any layer 3 backup interface for your peer links in case the peer link goes down. See below for more While a backup interface is optional, Cumulus Networks recommends you configure one. More information about configuring the backup link and understanding various redundancy scenarios is available below.
For example, if peerlink is the inter-chassis bond, and VLAN 4094 is the peer link VLAN, configure peerlink.4094 as follows:
Scaling this example out to a full rack, when planning for link failures, you need only allocate enough bandwidth to meet your site's strategy for handling failure scenarios. Imagine a full rack with 40 servers and two switches. You might plan for four to six servers to lose connectivity to a single switch and become single connected before you respond to the event. So expanding upon our previous example, if you have 40 hosts each with 20G of bandwidth dual-connected to the MLAG pair, you might allocate 20G to 30G of bandwidth to the peer link — which accounts for half of the single-connected bandwidth for four to six hosts.
To get a better understanding of how STP and LACP behave in response to various failover redundancy scenarios, read this knowledge base article.