[c-nsp] Layer 3 Core

[Shane Amante]

Another option to consider is, instead of enabling MPLS forwarding over 
your core, look at encapsulating Ethernet over MPLS inside L2TPv3.  This 
would require you to enable MPLS at the edge(s) of your network, but 
still run native IP forwarding in your core.

Overall, I would favor MPLS forwarding or L2TPv3/UDP/IP forwarding of 
EoMPLS packets between your locations because you get protection & 
restoration in the core, since that is a natural property of IP or MPLS. 
  In addition, you gain the benefit of fate-sharing and similar 
fail-over times for Layer-3 & Layer-2 connectivity between the two 
locations.  One downside of L2TPv3, to keep in mind, is it's not 
ubiquitously available in HW on all platforms, (e.g.: 6500/SUP720), 
whereas native MPLS is more widely available.

I think if you attempt to stand-up a single VLAN over the core, and you 
lose a link in the core, you're out-of-luck.  On the other hand, if you 
attempt to set-up diversely 'routed' VLAN(s) over the core, then you 
have to consider running STP between locations or start looking at other 
more complicated methods to provide protection between both locations.

-shane



Voll, Scott wrote:
> Why add complexity..... I'd just trunk.
> 
> Scott
> 
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Stephen Backholm
> Sent: Thursday, April 05, 2007 9:38 AM
> To: cisco-nsp at puck.nether.net
> Subject: [c-nsp] Layer 3 Core
> 
> Currently where I work we run a complete Layer 3 network core. This
> decision was made in order to keep Spanning Tree Protocol, ACls, QoS,
> and Policing out of the core and at the distribution layer. This has
> worked well for us, but we have a need for a few of our server VLANs to
> be in opposite geographic ends of campus for redundancy. These servers
> run as clusters and require Layer 2 connectivity between them, so in
> other words we need Layer 2 connectivity across our Layer 3 core.
> 
> 
> Here is the question.
> 
> Would it be better to solve this problem with MPLS or just trunk the
> handful of server VLANs across our core?
> 
> 
> Your thoughts and/or suggestions are appreciated.
> 
> Regards,
> Stephen Backholm
> 
> 
> _______________________________________________
> 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/
> 
> _______________________________________________
> 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/