Jon:
Here is the issue:
The client is largely a Juniper shop. And in the Juniper world, the routing engine (what we would call a SUP/RSP module) needs to have an IP address assigned and bound to it. This isn't an option, it's a requirement.
In the Cisco world, that is not necessary. We don't assign IP address to SUP/RSP modules. For management, we usually create a loopback - usually.
So, things cleared up after I got this understanding from the client regarding Juniper.
So, the client's concern - given their experience with Juniper - is how we handle failover in the event that the primary RSP fails and the secondary takes over.
Specifically, the question involves how we would manage the change of source IP address for the management traffic (aaa, ntp, ssh, etc) from the IP address assigned to the primary, to the IP address assigned to the secondary RSP? They want one management address to route and use for all management traffic.
[EDIT] It shoul dbe pointed out that, as a matter of policy, we are sourcing the RSP interfaces for management traffic [EDIT]
So, we even wondered if we could create an SVI for, say, vlan 10, make the switchports on the RSPs L2 ports, place them in the vlan, and the primary will co-opt that address. When it craps out and the secondary takes over, that one will co-opt the address. This way the same address is used.
My suggestion is to simply not configure any IP addresses to the uplink interfaces on the RSPs - we don't need them. Just shut them down for all we care. There's nothing special about these interfaces. For management, we can take the IP address presently assigned to the IP address on the primary RSP and apply it to any switchport on the 6748 blade, and make that the management interface. Then we shut down the interfaces on the RSPs (5/1, 5/2, 6/1, and 6/2), and be done with it.
What do you think?
