Re: [nsp] Trying to decide on IGP

From: Thomas Kernen (tkernen@deckpoint.ch)
Date: Wed Jul 29 1998 - 10:32:28 EDT


Hello,

would you have to enable the OSPF on the NAS itself or on the router
behind it (it there is a router behind). I know some NAS's (non cisco)
do have a hard time with OSPF.

Thomas

Niels Bakker wrote:
>
> Jon,
>
> I assume you want OSPF to be able to use redundant links and to allow
> customers with static IP addresses to dial into different POPs at
> different times while still keeping their IP address.
>
> > The options I'm considering are OSPF and iBGP. I think I'd rather go
> > with iBGP. I plan to continue to use OSPF on the home POP LAN, and will
> > probably use it at each remote POP, but I'm not sure if I want to try it
> > over WAN connections.
>
> Why not? It's an IGP, IGPs aren't limited to one location, only to one
> administration. You use BGP to announce yourself to the world
>
> > reflector. If I do run OSPF at each LAN, but use iBGP between POPs, then
> > I will probably have to carefully export OSPF into BGP.
>
> Why?!
>
> Here we have the following setup:
>
> - Network: cisco. At the POPs, we have a hub (or switch)
> connecting the cisco on our backbone to PM2s and PM3s, and
> possibly other hardware (customer leased lines etc.)
> - the ciscos generally run OSPF. The PM3s all do, the PM2s don't,
> mostly due to memory limitations (16 MB vs. 1 MB in the latest
> batch we ordered...). The ciscos have 'redistribute static
> subnets' under 'router ospf'.
>
> How do we use OSPF? We use it for customers with a static IP address -
> like batched SMTP and other business accounts (subnet routing). If a
> company logs in with an analogue modem on a PM2 we add a static route on
> the cisco with the PM2 as the next hop, else we let OSPF handle it by
> itself on the PM3.
>
> EuroNet, based in Amsterdam, The Netherlands, has multiple redundant links
> to the outside world. We're present at MAE-East and MAE-West and have an
> E3 to WorldCom (UUnet - not NLnet) in Amsterdam. Our backbone consists of
> a big fuckin' router at the center in Amsterdam, with most backbone
> connections - to the US, to UUnet, to various other Internet exchanges
> (DE-CIX in Germany, AMS-IX at the other end of the town, LINX in London,
> and via our office in Belgium to the BNIX), and the circuits to the POPs
> attached to it. All border routers are fully-meshed BGP speakers. All
> other routers speak OSPF.
>
> Yesterday we received a new VIP2 card and were able to move the
> connections to a few cities from asd-gw1 to asd-bfr with no
> reconfiguration except doing a 'no shutdown' and 'ip unnumbered Loopback0'
> on the interfaces and switching the cables.
>
> You don't outline what hardware you will have at the remote POPs. If you
> want some free advice: let cisco handle leased lines, and stick with
> whatever you have for dialup hardware.
>
> That way you can even whore yourself to cisco and get a free 'cisco
> partner / powered by' button to put on your homepage.
>
> Take care,
>
> --
> Niels Bakker, * * EuroNet Internet BV
> Network Operations * * Herengracht 208-214
> * 1016 BS Amsterdam
> NJB9 * +31 (0)20 535 5555






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