We have a 2pod ACI deployment with a VCF environment in each pod. The VCF environments connect to ACI with their own L3Out in a user tenant. We have a single, spanned, L3Out that interfaces with the rest of our enterprise. We desire to have the routes that are advertised from the external environments transit ACI and get re-advertised to the rest of the enterprise via both links. The trouble is that each pod has an associated firewall HA pair so we need to maintain return traffic state.
The L3 routing guide kind of looks like transit routes get a default tag of 4294967295 and it cannot be adjusted on a per L3Out basis. Does that sound right? Yes, to prevent route loops, fine, but we're all consenting adults here, to borrow a phrase, can't we just override that?
I'd LIKE to be able to tag routes advertised from each pod with a unique tag so I can adjust metrics as they're redistributed to the rest of the enterprise. Then I can force traffic back through the right firewall unless that path went away. Then, traffic could traverse the other firewall. I'm fine getting rid of the spanned L3Out if that's what it takes.
Thoughts?
