RE: Carrier supporting carrier: Why not L2VPN?

Cisco supports both VPLS BGP autodiscovery and H-VPLS. VPLS no doubt is worth
looking at for certain applications, but for a service provider using it for
long haul transport it might not be a good design option. As with anything
there are always both pros and cons, where a major pro of VPLS is that it's a
layer 2 extension as opposed to CSC which introduces more complexity in the
layer 3 routing design, but a major con of VPLS is that it's Ethernet only.

If an SP wants to use OC-192 or OC-768 POS as their long haul transport then
this won't work with VPLS. Another issue could be that if the SP buying
transit is using IS-IS for their routing, there are some failure situations of
VPLS pseudowires that IS-IS can't deal with because it doesn't have a
point-to-multipoint behavior like OSPF does.

Brian McGahan, CCIE #8593 (R&S/SP/Security)
bmcgahan_at_INE.com<mailto:bmcgahan_at_INE.com>

Internetwork Expert, Inc.
http://www.INE.com

From: Adam Booth [mailto:adam.booth_at_gmail.com]
Sent: Sunday, December 30, 2012 12:09 AM
To: Brian McGahan
Cc: John Neiberger; ccielab_at_groupstudy.com
Subject: Re: Carrier supporting carrier: Why not L2VPN?

Hi Brian,

I would agree that vanilla VPLS has scaling problems compared to other options
such as CSC but there are ways to enhance it to be worth looking at it. VPLS
using BGP Auto Discovery should overcome the control plane scaling and
provisioning headaches and using PBB with H-VPLS would overcome the forwarding
plane scaling limitations. I'm not so sure about Cisco (and would be
surprised if they didn't have the facility) but I know that Alcatel-Lucent and
I think Juniper support these methods.

Cheers,
Adam

On Sun, Dec 30, 2012 at 1:23 PM, Brian McGahan
<bmcgahan_at_ine.com<mailto:bmcgahan_at_ine.com>> wrote:
It is a corner case design. Most SPs would buy lambdas on someone else's DWDM
network or MPLS L2VPN etc. Routing at layer 3 through another SP
overcomplicates the design.

It's the same as like Inter-AS MPLS L3VPN option C. Just because you *can* do
it doesn't mean you *should* do it.

Personally I've never seen any SP run option C, and I've seen many many large
implementations. I'm sure it's out there and maybe someone here on the list
has seen it but the amount of politics and interoperability that's needed
between the providers is the limiting factor.

From a technology standpoint CSC and option C L3VPN are more scalable than
dark fiber or DWDM or L2VPN AToM or VPLS, but from an operational point of
view a lot of time they're just not practical.

Brian McGahan, CCIE #8593 (R&S/SP/Security)
bmcgahan_at_INE.com<mailto:bmcgahan_at_INE.com>

Internetwork Expert, Inc.
http://www.INE.com

On Dec 29, 2012, at 8:00 PM, "John Neiberger"
<jneiberger_at_gmail.com<mailto:jneiberger_at_gmail.com>> wrote:

> I'm watching some videos for the SP track and I just started one on CSC
> using L3VPN. It seems awfully complicated. I haven't even really gotten
> into the configuration of it yet, so maybe I'll be able to answer my own
> question shortly, but why not just use L2VPN for this? It seems like you
> could accomplish basically the same thing with MPLS pseudowires or
> something like that. Or maybe even VPLS or whatever. Not sure about that
> since I haven't done VPLS yet. :-)
>
> So, what's the deal? What advantage does CSC via L3VPN bring that would
> make it worth all the configuration hassle?
>
> Heck, why not just run MPLS over GRE?
>
> I'll keep watching the video and reading about it. There is probably some
> huge advantage that I am completely missing. I'm a noob to some of this, so
> go easy on me. :)
>
> John
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html

Blogs and organic groups at http://www.ccie.net
Received on Sun Dec 30 2012 - 16:39:55 ART