Re: [Sidrops] www.rpkiviews.org - geographically diverse vantage points

[Job Snijders <[e-mail]>]

Hi Tim,

On Tue, Jan 05, 2021 at 09:53:57AM +0100, Tim Bruijnzeels wrote:
> I think geographically diverse vantage points are indeed valuable.
> 
> However, a lot (most) publication points only have a single point of
> presence, so I suspect that a significant part of the variation in
> data seen is due to timing differences rather than geography / net
> topology.

Yes, timing also plays a big role in it all.

> Not a criticism.. just saying it would be good to keep this in mind
> when analysing differences.

Yeah, analysing differences is a tedious chore, there is an infinite
number of 'pathways' from signer to validator. Just looking at the 5 TAs
I observed the following:

    ARIN: 3 x A & 3 x AAAA DNS records for rpki.arin.net and also
          rrdp.arin.net, making for (at least) 12 points of presence.

    LACNIC: similar to ARIN

    RIPE: 1 x A + 1 x AAAA record for rpki.ripe.net, RRDP is distributed
          through a CDN with a 100+ points of presence.

    APNIC: one IPv4 address and separately one IPv6 address

    AFRINIC: similar to APNIC.

The above description of course is just the 'expected steady state',
Separately, there might be situations in which a publication point's IP
prefixes are hijacked in some way (be it via BGP or DNS trickery),
temporarily creating a 'multiple POP' situation.

In really complex cases one might even have to correlate routeviews with
rpkiviews, all the while keeping in mind that the two planes can
influence each other to some degree. Debugging the RPKI is hard work.

Kind regards,

Job