[c-nsp] BGP - Announcing routes to Internet providers.

Scott Granados gsgranados at comcast.net
Tue Jan 5 10:02:03 EST 2010


The memory impact isn't that bad and one person's over kill is another 
person's good planning ahead of time.  Why not do something right the first 
time and prevent the redesign / reconfiguration down the road which makes 
things that much more tricky in the long term.  I can't tell you how many 
messes I get dragged in to that need cleaning up because someone took the up 
front short cuts.  We're not talking about rocket science here, from the 
atlantic.net address and from Drew's long history on the list I assumed (and 
I think correctly) that there was the required clue there and justified 
need.


----- Original Message ----- 
From: "Ivan Pepelnjak" <ip at ioshints.info>
To: "'Scott Granados'" <gsgranados at comcast.net>; "'Drew Weaver'" 
<drew.weaver at thenap.com>; "'Cisco-nsp'" <cisco-nsp at puck.nether.net>
Sent: Monday, January 04, 2010 11:30 PM
Subject: RE: [c-nsp] BGP - Announcing routes to Internet providers.


Let's back a step and ask the questions we should have been asking in the 
first place:

* Are you an end-user or a Service Provider (somewhat reliable answer could 
be gleaned from Drew's e-mail address)?
* What's the size of your network?
* How many uplinks do you have?
* How far apart are your uplinks?

If it turns out Drew's uplinks are close together, all the beautiful design 
ideas presented here are a huge overkill.

And, BTW, I wish those of you that propose redistributing connected and 
static routes into BGP a huge budget you'll need to upgrade RAM and TCAM of 
your routers/switches when everyone decides (after reading this mailing list 
:) that following your recommendations unconditionally is a good idea :D

Ivan

> -----Original Message-----
> From: Scott Granados [mailto:gsgranados at comcast.net]
> Sent: Monday, January 04, 2010 10:03 PM
> To: Drew Weaver; Cisco-nsp
> Subject: Re: [c-nsp] BGP - Announcing routes to Internet providers.
>
> Drew, network statements are for the weak.:)
> (I'm kidding of course) but there is a better way.
> You should use community tagging in combination with prefix lists and
> route
> maps.  The idea is that you announce routes according to a tag and the
> behavior of the announcements depends on the specific tag applied.  For
> example, you could tag routes as peers, transits, global announce, etc and
> formulate the type of feeds you give your customers by filtering against
> communities so a customer wants peers and customers only you could match
> the
> two appropriate community tags.  This also allows you to tag the
> communities
> you globally announce uniquely and make the announcements in a unified way
> at your edges.  If you accompany this method with the appropriate
> redistribute static, redistribute connected, etc and use route maps to
> control this behavior you can remove the need for network statements
> completely and greatly decrease the things you need to modify and as a
> result the possible mistakes.  The other upside here is you can mark your
> more specifics as do not export and better control traffic internally
> better
> directing the traffic in your example.  It also allows you to accept
> communities from your customers and have automatic actions taken based on
> the tags they apply.  Let me know if you need some configuration examples.
>
>
>
> ----- Original Message -----
> From: "Drew Weaver" <drew.weaver at thenap.com>
> To: "Cisco-nsp" <cisco-nsp at puck.nether.net>
> Sent: Monday, January 04, 2010 12:35 PM
> Subject: [c-nsp] BGP - Announcing routes to Internet providers.
>
>
> > Howdy,
> >
> > I am trying to figure out if there is a different/newer/better(?) way to
> > announce our public IP ranges to our Internet providers, currently we
> are
> > declaring our subnets in 'network statements' in the BGP configuration,
> we
> > have static routes setup like ip route x.x.x.x 255.255.224.0 Null0 254
> and
> > then we have a extended access-list applied to each peer with our net
> > blocks listed in them.
> >
> > It appears that because of the network statements, the supernet routes
> > (/18s, /19s, etc) are being distributed via BGP to the rest of the
> network
> > which is by design(I assume). This doesn't seem ideal because if traffic
> > is sent to an IP address that doesn't have a more specific route than
> say
> > /18, or /19 it travels all the way through the network to the edge
> before
> > stopping. I might be blowing the impact of this out of proportion, but
> it
> > just seems like a waste of resources.
> >
> > Does anyone know of a seemingly more sensible way of doing this?
> >
> > -Drew
> >
> >
> >
> > _______________________________________________
> > 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/
>




More information about the cisco-nsp mailing list