Hi,
we have a georedundant dark-fiber connection between two company sites.
We're using a setup of 2 9336C-FX2 on each site, where each switch is directly linked to a single peer on the other site.
The 2 switches in each site are linked using vPC. We aggregate the two channels with LACP to make use of the redundancy.
Since the length of the actual fiber cables differs significantly, we do get almost twice the ping latency on the longer connection.
With our current setup, this however means, that depending on "seemingly random" factors (MAC/IP address) some machines get a 4ms and others a 8ms ping to the remote location.
Is there any way to change our setup to prioritize the faster connection as long as both links are up?
The link aggregation then should only switch to the slower connection if the faster connection is cut for some reason (somebody digging fiber cables out of the ground..)
In the ideal world, we would love to still "overflow" to the slower link, if the faster link is saturated.. but that's not as important as that isn't very likely to happen often.
Thanks for any feedback!
- Joachim
