-2

The bridge interface under Linux is quite dumb. In the end, it's just very inconvenient to use: you have to delete all IP configurations from the attached interfaces and somehow migrate the configuration to the bridge interface. I don't understand why (it doesn't make sense at all - normally, a bridge is not an endpoint for communication).

OpenBSD is much better; a bridge interface acts exactly like a physical switch. You create it, add the interfaces and you're done. Is it somehow possible to get similar behavior in Linux?

1
  • The problem you're facing is that BSD's bridge was designed to be a bridge, with a few switching features added later. Linux's bridge was designed to be a switch, but I'm not sure they ever really got there.
    – Chris S
    May 15, 2012 at 18:53

2 Answers 2

4

Bridges in Linux actually behave exactly like a physical switch.

Hey, let's take a look at my physical switch now. Can't get much switchier than a Cisco switch, right?

core-sw1#sh ip int brief
Interface              IP-Address      OK? Method Status                Protocol
Vlan1                  unassigned      YES NVRAM  up                    up      
Vlan2                  192.168.0.2     YES NVRAM  up                    up      
Vlan53                 10.53.0.2       YES NVRAM  up                    up      

Hmm… on my Cisco switch the IP addresses are actually assigned to… the bridge interface. Not a single port.

Yep, you're thinking about it wrong. Consider what an IP address assigned to a single interface on a bridge would mean.

1

you can use openvswitch in bridge mode.

Not the answer you're looking for? Browse other questions tagged .