Switching

Greetings.

We're building a new datacenter using all Juniper gear, and coming from a Cisco background, we're curious about multilayer switch / L3 interface redundancy.

In Cisco, we'd build two multilayer core switches, and "cluster" the two using an HSRP VIP. Of course, you can create a similar configuration in Junos with VRRP using two cores, but what if your core is a virtual chassis?

Essentially, I'm going to build a multilayer core using a virtual chassis of two 4500s and a few 4200s as "line card" role switches. If I create an RVI on one of the 4500 switches (say member0/RE0) and that switch dies, will my RVI automagically "float" down to the next 4500 or must I still create a VRRP pair across two of the 4500s?

Apologies if this is in the docs somewhere already... I just haven't run across it yet.

Thanks in advance.

That's an excellent question. The answer is that you don't need VRRP. Think of that Virtual Chassis like you would of a physical chassis, with the two EX4500s being your "SUPs" - "REs" (Routing Engines) in Juniper speak. The EX4200 are your line cards. When the main RE (1st EX4500) fails, the backup RE (2nd EX4500) will take over, with all configuration (including L3 interfaces) intact.

As an aside, you can do this with a small core using EX4500/4200 (or even just 4200), and you can also do it with a large core using two EX8200 physical chassis in one virtual chassis. Put that in your pipe and smoke it. 🙂

That's pretty sweet. Removing the complexity of additional IP addresses (real IPs + VIPs) and redundancy protocols will make life easier, for sure.

Thanks for the reply.