Ok, so, at my work site, we use tagged VLANs. We're also starting to roll some physical Linux servers (all prior rollouts of Linux had been as VMs). It seems there's two ways to combin VLAN tagging and bonding:
- Bond physical interfaces, then tag the bonds
- Tag physical interfaces, then bond the vnics
In practice, is there any real difference between the two ways of layering the protocols? Pros and/or cons to each method? I've not really dug up any "best practices" on the subject, so, wanted to see what others have chosen to do and why.