Forum Discussion
If I have ports available, I usually dedicate a 'peernet' link (1gbps has always been fine for me) for f5-to-f5 connectivity, solely for the purpose of config-sync, failover, and mirroring. I then use HA-groups to monitor my production links (NOT the peernet link) to provide network-aware failover in the event production links fail. I always want my F5's to see each other and be able to communicate failover information regardless of connectivity to the rest of the network. If I don't have spare ports, I would do as Arnaud mentioned - tag a peernet VLAN across the production LACP trunk.
I haven't seen a situation where I ever needed excessive capacity for the peernet link, but I've also not actively monitored usage of that link. The F5s are only transferring state information (failover state, HA scores), config-sync data (at most a full copy of the current local UCS), and mirrored data (connection table). I think you would be having other device/capacity issues well before you saturate a link with peer-to-peer communication