[c-nsp] L2TPv3 vs. VPLS

Andrew Fort afort at choqolat.org
Mon Sep 19 18:57:20 EDT 2005


On 19/09/2005, at 7:02 PM, Vincent De Keyzer wrote:

> Jeff,
>
> in short, what are those major limitations?
>
> Vincent
>

with VPLS, mostly issues of maturity and complexity (hence H-VPLS and  
the like, though arguably this makes things more complex, not less).   
what happens if you need to split your VPLS domain up and police/ 
control individual VCs differently over different parts of your  
network (e.g. lower rates on your WAN versus your own metro fibre  
network), some vendors don't have all the knobs you might like, on  
the platforms you might be using.

at present, it would seem that if you're seriously considering VPLS,  
you're looking at vendors other than cisco.

as most are already aware, you really only need VPLS if you have a  
legacy l2 network built from ethernet switching gear entirely and  
you're already offering P2MP services.  even then, depending on the  
size and growth potential of the network, it may be suitable to use  
P2P psuedowires + traditional VLAN switching (particularly on cisco  
kit at this point).

Virtualisation at the switch level is useful (the 'virtual switch  
instance' concept) so that each VPN gets its own VLAN tag space (much  
like a VRF routing table), and some hardware switching platforms  
(e.g. c7600), at least as i understand it, don't use a 'large'  
internal VLAN tag, so cannot do this (hence 'internal' VLANs eating  
up your 12-bit space).

riverstone and alcatel apparently have some boxes worth considering  
in the VPLS/L2VPN area.

-andrew

>
>
>> If you don't need P2MP and don't have MPLS backbone in place don't  
>> even
>> consider VPLS, there are still some major limitations.
>> I think either L2TPv3 on layer 3 or QinQ on layer 2 would be  
>> suitable for
>> you. In the past I used L2TPv3 between 2x10720 to transport 4Gb L2  
>> traffic
>> between 2 IX's, worked just fine, the only issue we've had was MTU  
>> when
>> main
>> STM16 ring went down and L2TP tunnel was rerouted via Ethernet links.
>>
>> Hope this helps,
>>
>>
>
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>



More information about the cisco-nsp mailing list