[c-nsp] MPLS best practices question

[Mark Tinka]

On Tuesday 29 June 2010 03:10:07 am Christopher E. Brown 
wrote:

> That only works until aq high $$$ customer starts
>  demanding 9000byte payloads for their IP in vrf or VPLS
>  service...

We run two networks - one for IP Transit, another for 
national Metro-E (different companies, so don't think "NGN" 
or any of that marketing, hehe).

We use 9,000 bytes for the IP Transit network, since the 
majority of our customers come in at 1,500 bytes (even 
though those ports run a higher MTU value), and the other 
end of their interest is also coming in at a typical 1,500 
bytes (transit, peering hand-off, e.t.c.).

On our Metro-E network, we've been running at 9,216 bytes, 
mostly because that's over 80% Juniper-based. The 3750ME's 
only talk 9,000 bytes (and yes, some customers have 
requested for even 9,200 bytes as payload - go figure). So 
we've been let down by the platform, but the edge has tended 
to dictate the direction.

More recent platforms that we're deploying will be 
supporting higher MTU values in the edge, and not having to 
reconfigure the core is a plus. It just wasn't a problem in 
the past because IS-IS didn't need to extend into the 
access. Now it does, and we always want to avoid any 
trickery when abstracting MTU values between disparate links 
re: adjacency formations. The 9,000 bytes in our IP Transit 
network allowed us that uniformity, given the network's 
function and limitations in the access.

Cheers,

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20100629/ea603e5f/attachment.bin>