Bridge-domain vs xconnect

[xzjleo2005]

Hi guys, I have been readying a few documents about VPLS/EoMPLS but still confuse about the bridge-domain and xconnect. Could you please provide any ideas which scenarios I should use bridge-domain and which should be for xconnect? What is the difference between them, any documens can explain this?

Thanks, Leo

[xthuijs · Accepted solution]

The simple difference between the 2 is mac learning.

An Xcon will just throw everything it received over to the other end.

A bridge-domain will forward traffic based on the dmac knowing where it should go. If it doesnt know it goes flooding.

So if you have 2 circuits to connect only, VPWS or XCON is the right choice as it is simple, light weight and fast.

If you have more then 2 end points you will need a Bridge Domain which constitutes mac learning with the notion that flooding is intensive from a hw forwarding perspective and will consume more system resources in terms of mac tables.

xander

---

Xander Thuijs #6775

Principal Engineer ASR9000

[xzjleo2005]

Very clear, thanks Xander.

[Ugur Ersoy]

hi alexander, thank you for this clear info, yet i need to know more about vpls, in particular bridge domain, service instance, and some terms like rewriting, popping, pushing. can you please suggest me a source? i was not able to find any.

thanks in advance.

[xthuijs]

hi Uger, this may be a good ref for that:

https://supportforums.cisco.com/document/59741/asr9000xr-flexible-vlan-matching-evc-vlan-tag-rewriting-irbbvi-and-defining-l2

Also in Cisco Live 2013 Orlando session ID 2904 I covered some of that for more detail if you like.

cheers!

xander

[Ugur Ersoy]

thanks Xander, this page is what i am looking for start and becoming familiar with terms.

[donalddangelo]

Xander,

You mention MAC learning using bridge domain, is there a way to view this MAC table?

Edit: I managed to figure out how to view the bridge domain table "show l2vpn forwarding bridge-domain mac-address location 0/0/cpu0" in my case.

[xthuijs]

ah you found the command already. that is correct:

show l2vpn forwarding bridge-domain mac-address loc  0/7/CPU0

will show you the mac table.

also have a look at this discussion which might be interesting for you:

https://supportforums.cisco.com/discussion/12501141/9k-view-mac-address-table-question

xander

[andresfcl]

 

Sorry to reopen these but i have a doubt regarding bridge domain operation.

mutiples access pseudowire directly under a bridge domain

Does any body could help me to understand the difference between using a using a vfi instead of having just multiples access pseudowires under same bridge domain, is there any difference.

I am reaaly intrigue of how it works when i have multiple neighbor configured under same bridge-domain. I tough that if we need to connect to  multiple neighbors  from a  bridge-domain  we always have to use vfi.

Bellow are the examples I came accors..

bridge-domain gestionxxxxx
mtu 9000

interface TenGigE0/3/0/2.40xx
interface GigabitEthernet0/1/0/4.10xx
!
neighbor 200.xx.254.31 pw-id 10xx
!
neighbor 200.xx.254.32 pw-id 10xx
!
neighbor 200.xx.254.102 pw-id 10xx
!

What si differece if i have instead have used a vfi to put eachneigh under the VFI config.

Whats the loging if a unknow unicast or boeadcast traffic come from onw of the AC or one of the pseudowires ??????? 

[andresfcl]

Does any body could help me to understand the difference between using a using a vfi instead of having just multiples access pseudowires under same bridge domain, is there any difference.

I am reaaly intrigue of how it works when i have multiple neighbor configured under same bridge-domain. I tough that if we need to connect to  multiple neighbors  from a  bridge-domain  we always have to use vfi.

Bellow are the examples I came accors..

bridge-domain gestionxxxxx
mtu 9000

interface TenGigE0/3/0/2.40xx
interface GigabitEthernet0/1/0/4.10xx
!
neighbor 200.xx.254.31 pw-id 10xx
!
neighbor 200.xx.254.32 pw-id 10xx
!
neighbor 200.xx.254.102 pw-id 10xx
!

What si differece if i have instead have used a vfi to put eachneigh under the VFI config.

Whats the loging if a unknow unicast or boeadcast traffic come from onw of the AC or one of the pseudowires ??????? 

[Aleksandar Vidakovic]

All PWs you put into a VFI will belong to a split horizon group. BUM packets (broadcast, unknown unicast and multicast) frames received from one member of the VFI are never forwarded to other members of the VFI.

/Aleksandar

[vladdar1349]

Hello, I have another dumb questions regarding this. I have a simple setup when I am using xconnect to bridge one AC (attachment circuit) from customer with one primary and one backup PW neighbor. Everything works fine until the customer starts pushing multicast traffic (with unicast it works fine). Looks like the PW neighbors (which are older ASR 1002) are not able to handle the BUM traffic and LDP sessions start flapping.

ASR9k

!

xconnect group custX
p2p custX
interface GigabitEthernet0/3/0/3
neighbor ipv4 x.x.x.x pw-id 496
backup neighbor x.x.x.x pw-id 496

Which config should I use in this scenario?

Thanks.

[Aleksandar Vidakovic]

Are you saying that the asr9k is forwarding traffic both into the active and backup pseudowire? This should not happen (and I very much doubt it can happen on asr9k). Or are you saying that the asr9k is receiving traffic on the backup pseudowire? But even if that happens, it shouldn't lead to LDP flaps. It would be good if you could clarify the problem a bit more.

/Aleksandar

[vladdar1349]

Hi Aleksandar,

thanks for replying. AC is on asr9k and multicast is forwarded through the pseudowire to the neighbor asr1002. I don't think that asr9k is sending the traffic to both neighbors but I was seeing higher CPU utilization on primary neighbor asr1k and then LDP/OSPF flaps towards this neighbor. I suspect after asr9k lost LDP peering with primary neighbor it switched to the backup neighbor and the situation repeated itself on the backup neighbor(which si also asr1002) and then over and over again. Multicast traffic from customer seems to be the reason for it, after stopping this multicast traffic everything returned to the normal. I am just wondering if this config is valid for multicast service or why the asr1002 can't handle it...

[Aleksandar Vidakovic]

p2p xconnect on asr9k really doesn't care about the traffic type because there's no MAC learning. What comes in on one end goes out on the other. :) This definitely needs to be investigated on the asr1k side. I'm afraid I can't help much with that because it's not an IOS XR platform.

Btw, if you want to slow down the switching back from backup PW to primary after the primary PW comes back up, you can make use of the "l2vpn pw-class backup disable delay " command.

/Aleks