Re: [nsp] Newbie OSPF/RIP question...

From: John-David Childs (jdc@nterprise.net)
Date: Fri Dec 12 1997 - 15:53:40 EST


On Friday December 12, 1997, Bruce R. Babcock <bbabcock@cisco.com>
 had this to say about "Re: [nsp] Newbie OSPF/RIP question...":

> If you don't want to assign an address/sub-net to the WAN link to the
> upstream provider you might look into PPP with IP unnumbered. You should
> find a good description on our web page.

Thanks for your reply! I have questioned why they don't use PPP
unnumbered or static routing, but they like to charge us mere mortals any
time we give them a call "if it's not their problem".

It would seem that I could just inject a route statement into IGRP saying
"yes, I'm always responsible for the following aggregate routes (the four
/24's we've been allocated)" but I'm not quite sure how to do it. Anytime
I remove (what I call a "hold-down" IP) from the loopback interface, the
route for that entire /24 dies at the upstream (it works fine internally).

I am redistributing my OSPF process into (their) IGRP, and I have network
commands for each of my /24's in the OSPF process. It all works fine
internally.

> IP unnumbered off a loopback interface (single host address) is normally used.
> If you want a numbered interface, maybe the upstream provider might be
> willing to allocate from his IP address space using a /30 mask.
> The usual way to support numbered interfaces and still conserve IP address
> space is to run OSPF which supports VLSM and use /30's for WAN links.
>

Yup...this is what I do with one of the other companies I sysadmin and
their upstream. Works great.

> With only a single provider, running -any- IGP with your provider could be
> very unwise. Static routes are preferred.
> Even with the best of access lists, eventually someone will forget to
> maintain the lists and you could end up trashing each other's IGP's and
> perhaps networks. I've seen it happen a lot of times. Any instability on
> either end impacts the other in the form of more wasted WAN bandwidth and
> increased CPU to process the route flaps.
> Redistribution also gets messy from one IGP to another.
>

Oh, goody...maybe I'll assign myself one of their /24's and see what
happens...that'll get 'em to change their routing policy :-) :-) Thanks
for your reply.

--
John-David Childs (JC612)       Enterprise Internet Solutions
System Administrator            @denver.net/Internet-Coach/@ronan.net
  & Network Engineer            1031 S. Parker Rd. #I-8 Denver, CO 80231
As of this^H^H^H^H next week, passwords will be entered in Morse code.



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:14 EDT