[arin-ppml] Rationale for /22

[Jon Lewis]

On Tue, 28 Jul 2009, William Herrin wrote:

> On Tue, Jul 28, 2009 at 2:09 PM, Jon Lewis<jlewis at lewis.org> wrote:
>> The distinction some people may not be getting is that if I know ARIN
>> allocates from a /8 nothing longer than /20s, then if I'm running out of
>> routing slots, I can use a prefix-list to ignore anything /21 (or maybe /22)
>> or longer from that /8.  If ARIN allocates /24s from a /8 or probably longer
>> net, then I need to accept those /24s.  That's the theory anyway.
>
> Hi Jon,
>
> That theory was destroyed by NRPM 4.2.3.6.

https://www.arin.net/policy/nrpm.html#four236

How so?  Suppose you're multihomed to Sprint and XO and are using a /24 
from XO space.  Lets say I'm on two other networks and I filter on RIR 
minimums.  I normally get to you via an XO aggregate route (say a /17 that 
covers your /24).

Your XO circuit goes down.  I still send traffic destined for your /24 via 
the path I see for the /17.  A few things can happen.  The traffic makes 
it as far as XO, at which point they hopefully are accepting your /24 via 
your advertisement to Sprint, and the traffic gets to you (it's a longer 
path than usual...but at least it made it).  I send traffic for your /24 
via the path I see for the /17, and before it gets as far as XO, a router 
in another AS along the way has "full routes" and has a path for your /24 
via Sprint and the traffic gets to you about as efficiently as normal.

What broke?

The breakage happens when people deaggregate PI space (no covering 
aggregate announced by an upstream) and don't announce the aggregate.  I 
have to assume these are cases of stupidity/negligence...but there are 
many examples of it.

----------------------------------------------------------------------
  Jon Lewis                   |  I route
  Senior Network Engineer     |  therefore you are
  Atlantic Net                |
_________ http://www.lewis.org/~jlewis/pgp for PGP public key_________