[j-nsp] full table?

[Dan Farrell]

My philosophy on table size and usefullness is tied to the AS distance of said routes.

Arbitrarily speaking, any destinations more than 5 AS Hops away isn't likely to need the most efficient BGP-provided route to hit the destination. At that point, 6 actual autonomous systems is basically 'across the internet', and if you have 2 or 3 (or more) providers, they likely normally take the same relative ip hops and latency to reach the same destination when compared to each other. For example, if you are a domestic US network, getting to a network in Malaysia via Sprint or via AT+T (given ok network conditions and these carriers do not directly serve the destination network) will usually have the same relative latency and hops.

It's the destinations that are 1 to 4 Autonomous Systems away where BGP can make a palpable difference. For example, you don't want to transit Sprint to AT+T to an AT+T customer if you already peer with AT+T. Keeping both of those routes would be more important than keeping the two Malaysian route destinations in the previous example. And with limited memory, you might be in the position to have to make that choice.

So if you limit your route tables to a set number of autonomous systems (AS hops) then you get the 'relevant' routes, and all of those on the periphery can be be a 'carrier-grab-default' or a least-cost link. Also, those operators who love to as-path-prepend the hell out of a route will find that their routes don't even make it into your tables (which is kind of the point).

I always ask for full tables and the default from each provider, and then I hone down from there as necessary.


Dan Farrell
Applied Innovations
danno at appliedi.net
________________________________________
From: juniper-nsp-bounces at puck.nether.net [juniper-nsp-bounces at puck.nether.net] On Behalf Of Keegan Holley [keegan.holley at sungard.com]
Sent: Tuesday, September 20, 2011 1:26 PM
To: juniper-nsp
Subject: [j-nsp] full table?

Is it always necessary to take in a full table?  Why or why not?  In light
of the Saudi Telekom fiasco I'm curious what others thing.  This question is
understandably subjective.  We have datacenters with no more than three
upstreams.  We would obviously have to have a few copies of the table for
customers that want to receive it from us, but I'm curious if it is still
necessary to have a full table advertised from every peering.  Several ISP's
will allow you to filter everything longer than say /20 and then receive a
default.  Just curious what others think and if anyone is doing this.
_______________________________________________
juniper-nsp mailing list juniper-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp