Last Mile Design

Brandon Martin lists.nanog at
Fri Feb 8 00:37:32 UTC 2019

On 2/7/19 6:46 PM, David Ratkay wrote:
> I am not sure if this is a easy question to answer. But I am wondering 
> what ISP's do for their residential and business customers for designing 
> POP's that they usually access to get theur traffic into a given ISP and 
> beyond. Is it usually a L1/L2 connection from the CE to the last mile 
> POP? Or L2 even within the last mile POP. Do you just have POP's 
> delegated to residential users and a separate POP for business users. Or 
> is it done on a geographical basis. So for this region of City-A we 
> manage both residential and business customers at this same POP.

L3 switches that can handle a reasonable number of routes/VLANs/MACs and 
lots of bandwidth are so cheap that I'm fond of pushing L3 fairly deep 
into the access network with them in many cases.  Not much benefit to 
that if you prefer centralized BRAS/BNG style boxes with all the bells 
and whistles to take the traffic management away from your last-mile 
gear, though.  So you need access gear with its own traffic management 
capabilities and potentially L2 filtering of higher level traffic (DHCP 
snooping, ARP/ND inspection, RA guard, TCP/UDP port blocking, etc.) and 
that may limit your options or force you to terminate fewer customers at 
a PoP than you'd like to stay within the capabilities of a typical L3 
switch product.

I've never been overly fond of the Ma' Bell style designs with humongous 
routers in centralized areas and L2-only haul out to the last-mile 
termination.  The failure modes of such systems often result in 
hilariously large outages that are super visible publicly and put egg on 
peoples' face.  A neighborhood being down is a little easier to manage 
from a customer relations POV, I think, and it's easy to make that 
happen with distributed L3 termination.

I've also found it easier to handle multiple backhaul paths at L3 than 
L2 since spanning tree is such a pain in the butt, but E-RPS/G.8032, if 
you get switches that support it, can also be very handy.

There are some smaller, somewhat cost-effective full-touch routers that 
can help bridge the gap between those two options, though.  Juniper's 
MX104 and the Cisco ASR1k series are some reasonable options for that, 
but it'll definitely cost more than a cheap L3 switch for a given amount 
of bandwidth.

I do like to separate SMB and Resi traffic, but it's mostly for customer 
service reasons rather than technical reasons.  That separation rarely 
entails separate equipment but rather just VLANs and PCPs, IP subnets, etc.

Now if you want to sell DIA type services where you can offer full BGP 
tables, MPLS interconnection, etc., that's another matter.  A need for 
IPv4 CGNAT may, as well, but things like 464XLAT, lw4o6, MAP, etc. can 
fix that up if you're willing to put some extra requirements on your CPE/RG.

If you're in a position where you want to or have to offer competitors 
access to your network to sell service directly to customers, that's 
also going to potentially really change the situation.
Brandon Martin

More information about the NANOG mailing list