nested prefixes in Internet
mysidia at gmail.com
Tue Oct 11 09:09:00 UTC 2016
On Mon, Oct 10, 2016 at 12:24 PM, Niels Bakker <niels=nanog at bakker.net> wrote:
> * r.engehausen at gmail.com (Roy) [Mon 10 Oct 2016, 19:19 CEST]: [snip]
>> If the ISP-A/ISP-B link goes down then the /24 would be seen only via
>> ISP-C which is the desired result.
> What if ISP-A then receives traffic inside its /19 destined for ISP-B's /24?
> It will have to send it over transit and won't bill ISP-B for that traffic.
This is still the favored approach over ISP-A de-aggregating their
address space to
support some multi-homed customers.
It's all something that ISP-A can put in their contract with ISP-B.
A can bill ISP-B per packet-byte routed with destination of B's
/24, if they wanted,
or reach other mutually agreeable conditions that make sure the /24 carve out
does not commit A to giving service to B involving more traffic than B pays for.
Or they can list Rate #1 for incoming traffic to B's /24 sent to a
peering link with ISP-B,
and apply pricing Rate #2 for packets destined for B's /24
between two transit providers.
...However, ISP-A and ISP-B should have a link; Either a physical connection
or a virtual one (such as a tunnel), with BGP peering between A and B,
for ISP-B to be
using a /24 from A's IP space with other providers.
> You cannot expect 100% of the rest of the Internet to honour the more
> specific all the time.
That is true, but not really notable / isn't really a significant problem.
Probably close enough to 100% of the internet to honor it 99% of the time,
and the 1% that don't probably will likely not be one of ISP A or
B's adjacent providers.
(If they are, then A or B will open a ticket, and the /24 will be
fixed to work as desired)
Heck.... on some occasion when ISP A has an outage, the /19 won't be there,
so less than 100% of the internet might 'honor' that one at times, as well.
More information about the NANOG