Log in to ask questions, share your expertise, or stay connected to content you value. Don’t have a login? Learn how to become a member.
We are migrating from a pair of old Cisco Nexus switches to a pair of Juniper QFX5120 switches.
The Cisco switches operate independently, the Juniper QFX switches are in a virtual chassis.
The Cisco Nexus switches use Cisco HSRP for the high availability of our network gateways. We have self-IPs on each Cisco switch and then a floating/virtual IP which is the actual network gateway. The virtual IP then fails over from one switch to the other if anything happens to a single switch.
My understanding of how this would work with Juniper (if we had stand-alone switches) is that we set up VRRP with IRBs.
However, as mentioned, our two Juniper QFX5120 are in a virtual chassis.
Because they're in a virtual chassis, does that mean we don't need VRRP? The virtual chassis basically takes care of the high availability of the network gateways for us, and we just need to set up IRBs and the virtual chassis does its magic to take care of high availability for us?
Also, with the Ciscos, we had 10 network gateways, and we put five of those on one Cisco switch and five of those on the other to distribute traffic evenly.
Does the Juniper virtual chassis sort of take care of the traffic distribution of the IRBs for us?
With the irb virtual interface this will be active as part of the routing engine of the virtual chassis. So the ip would be present even if one of the member switches were to fail in a virtual chassis.
With the virtual chassis two members are elected as the primary and backup routing engine. So even if one of these fails the configuration and irb interfaces survive. This is roughly equivalent to an old style physical chassis with blades.