On Mon, Jul 29, 2002 at 01:21:56PM +0200, Pascal Gloor wrote:
> well, it seems some of you think filtering 30% of the routes is good...
> 

30% of what?? The filter I sent out contains only prefixes which never ever
should show up in the public internet and thus would filter 0%. OTOH filtering
100% of RFC1918 is a good thing.

> so, why dont you filter all the incomming routes(not peerings) and just
> load-balance your outgoing traffic?
> that will also work... but is that really what we want?
> 

though it may work, this is already broken by design.

>                                                                       I'm
> really interested to see the benefits of that...
> 

Have a look at the NANOG archives. This thread pops up every
few months (in the moment going on again).

To put it in a nutshell:

1. while not only table size is a matter, the more prefixes you have the
   more flaps/withraws you see. And also the shorter a prefix the less you see
   flapping. And flapping is what really hurts. So less flapping/withdrwas
   means great mprovement in stabilty.

2. address space which is still unallocated or private address sapce or special
   address space should neither show up in the global routing table nor at any
   IXP. I guess that goes without saying.

3. When it comes to announcing more specifics you open up a religious war.
   So I'll stop to say more about that.


Regards, Arnold
-- 
Arnold Nipper / nIPper consulting          mailto:  [EMAIL PROTECTED]
Hermann-Loens-Weg 15                       Phone:  +49 6224 9259 299
D-69207 Sandhausen                         Mobile: +49 172  2650 958
Germany                                    Fax:    +49 6224 9259 333
----------------------------------------------
[EMAIL PROTECTED] Maillist-Archive:
http://www.mail-archive.com/swinog%40swinog.ch/

Reply via email to