Bridged point to point T1 setup questions.

[amcsystemstech]

One of our locations is adding a small five user office to its business. They are going to connect the remote office via point to point T1. I've decided to use bridging to lessen the impact on the rest of our network.

I have a test setup up and working. DHCP even works accross the link. I just used no ip routing and assigned both serial and ethernet interfaces to the same bridge group. I also added a loopback interface and assigned the same ip address to loopback, ethernet and serial interfaces on the router (I assigned a different ip address to each router.). This allows me to access the routers remotely. Does this sound correct?

Also, running tcpdump on the test setup yields a high number of 802.1d messages. The messages list a pathcost of 651. Looking on the net, I didn't see anyone getting that high of a pathcost. Does this indicate a potential problem or misconfiguration?

Thanks,

[ilya.varlashkin]

Pathcost value derived from interface speed, higher speed - smaller value. T1 speed is smaller than Ethernet, that's why value is higher than you usually see. 802.1D specification has table with recommended values, however many vendors use different values.

If you've decided for bridged setup, then it might have been more correctly to configure management IP on BVI interface and not the same IP on loopback, ethernet and serial.

It might look like bridging is easier to setup and maintain, but you should take into account that WAN link speed is noticible smaller and having bridged environment all broadcast/multicast traffic (ARP, OSPF, DHCP, STP etc.) will travel across WAN link even if there are no receipients on the other side. This is inefficient use of network resources. Routing on the other hand separates broadcast domains and more suitable for WAN environments.

These days it's possible to get virtually any application working in routed environment, even those which rely on broadcasts. Because of this there basically no reason to configure "flat earth" network across WAN.

[amcsystemstech]

Thank you for taking the time to respond.

To use the BVI interface, do I have to enable irb or is it just as simple as adding the interface and removing the ip address from the other interfaces?

The reason I'm using bridging is because I don't manage part of this network. I was asked for a quick delivery with an absolute minimal cost. If I were to use routing, it would require action from one, maybe two other technology companies. This would add both cost and time to the project. Not the best of reasons, but the only plan of attack I could come up with.

Thank you,

[ilya.varlashkin]

You'll have to enable IRB, but that' only one line. Here is part of the config that might be close to your scenario:

bridge irb

bridge 1 protocol ieee

!

interface FastEthernet 0/0

no ip address

bridge-group 1

!

interface Serial0

no ip address

bridge-group 1

!

interface BVI 1

ip address

That's all. Note that BVI interface number must be the same as number of the bridge group.

[gpulos]

the configuration you've mentioned sounds correct for a bridged T1 WAN connection.

the 802.1d messages you are seeing are BPDUs. these are used by spanningTree and are, by default, generated every two seconds and propagated througout the LAN.

the cost of 651 sounds correct for a T1 point-to-point connection.

see the following link for STP cost calculation methods:

http://www.cisco.com/en/US/products/hw/switches/ps700/products_configuration_guide_chapter09186a008007fa34.html#wp1139504

fyi, for your internal network, you should never see path costs this high as you probably have at least 10Mb or 100Mb links between hosts/devices.