IXP

Nuno Vieira - nfsi telecom nuno.vieira at nfsi.pt
Sat Apr 18 08:23:30 UTC 2009


----- "kris foster" <kris.foster at gmail.com> wrote:

> painfully, with multiple circuits into the IX :) I'm not advocating  
> Paul's suggestion at all here
> 
> Kris

Totally agree with you Kris.

For the IX scenario (or at least looking in a Public way) it seems Another Terrible Mistake to me.

IMHO, when you are in a Public IX, you usually want to reach everyone's network without hassling around.  Then it is your problem, and yours peer problem if we peer or not.  

When you overload a certain port at a Public IX, you rather upgrade that Port, or, Move particular bit pushers and movers for a Private Peering port (if it really makes technical and economical sense).

I don't see how this idea that came out there could benefit the operational daily works (For IX, For IX Customers) , also, it would require work from the (usually) Neutral IX, when users need to connect ear other, which, will lead in more money to pay.  (hey IX OPS.. we are company X and Z, and we signed a nice peering agreement.. can you please virtual patch us ?)  Where is the neutrality here ? Time ?  What if my equipment brokes at 3 AM and IX Ops need to change configs ?  

Ok, ones could say... it is automated...  BUT.. what is the really security behind automation ? The portal is on the Wild Web, right ?   

This happens today on datacenters, with real cross connects, usually thru MMR's (Meet me Rooms).    I don't want to have a Virtual Meet me Room, on Internet exchanges where i peer. 

This is my view.  I might be wrong, but i don't care, as i am square as a rock. :-)

I don't understand how can this new concept (or really old, considering ancient ATM peering and stuff), can be better, more secure, and cheaper for all.

cheers,
--nvieira




More information about the NANOG mailing list