---

On Wed, Mar 20, 2019 at 11:01:30AM +0300, Andrey Korolyov wrote:
> >
> >
> > And when everything is made clear, if a report is filed against AS1, AS1's
> > holder might have a problem, so i see a strong reason for not even trying
> > :-)
> >
> >
> Out of interest, take an AS1 with single malicious upstream AS2, what stops
> AS2 to pretend that AS1 has made bogus announcements and make them for its
> own purposes? This situation looks pretty real without RPKI or other
> advertisement strengthening methods, as I could see. How experts are
> supposed to behave in this situation?

This has been seen many times, even chain situations like

<upstreams and peers> - AS X
\
AS 3 - AS 2 - AS 1
/
<upstreams and peers> - AS Y

where X and Y are legitimate ISPs, while {1,2,3} is basically a single rogue
entity - or a set of rogue entities closely working together with a common
criminal goal.

In such a setup, AS 1 should be considered as the most "throw-away" resource,
while AS 3 would play the "customer of customer, not my business" role,
and AS 2 would play the "i notified my customer and will disconnect them
if they continue" role. When AS 1 is burnt, a new one is made - with
new people as contacts, new IP addresses, etc, so that no obvious correlation
can be made. Most of the bad guys infrastructure is in AS 3 and that remains
pretty stable because their bad nature can not be easily demonstrated.

Whatever set of rules is made against hijacking, it should be assumed that
these groups will do everything to get around those rules, and many AS's
can be used to this end. Since there is no shortage of AS numbers, I
assume that anybody can get one easily so they can change them as if they
were underwear.

And yes, unallocated AS's in the AS 1 position, announcing unallocated IPs,
have also been seen. Those are even easier to get :-)

So the ideal scheme to counteract BGP hijacking should be able to climb up
the BGP tree in some way, until "real" ISPs are reached.

Nice discussion!

furio ercolessi