Suppose you have a set of Ethernet ports (say 1-4), and you want each of them to bridge both tagged and untagged traffic. The most straightforward way is to simply bridge the Ethernet ports.
But now suppose this device is the DHCP server for ten particular VLANs. Each DHCP server needs an explicit VLAN interface. I can think of two ways to do that:
- Define VLAN interfaces on the bridge, and associate each DHCP server with the corresponding VLAN interface.
- Define separate VLAN interfaces on each port, bridge those, and associate each DHCP server with the corresponding bridge.
Approach 1) requires 10 VLAN interface definitions (one for each VLAN being served) and one bridge (of the four Ethernet ports);
Approach 2) requires 40 VLAN interface definitions (one for each VLAN on each port), and 11 bridges (one for each VLAN, plus one for the ports themselves for untagged traffic)
My questions are:
- How will these two approaches differ in operation, and
- Assuming all VLANs are used on all four Ethernet ports, are there any advantages to the second approach that justify defining this large number of interfaces?