Maybe we're looking at the wrong place when dealing with TCP amp. I believe there is a much easier way to solve this.

@OP: can you post the tcp flags of the SYN/CK you are receiving from Sony?

Thanks

Jean

On Mon, Jan 27, 2020 at 5:43 PM Töma Gavrichenkov <ximaera@gmail.com> wrote:

On Tue, Jan 28, 2020, 4:32 AM Damian Menscher <damian@google.com> wrote:

On Mon, Jan 27, 2020 at 5:10 PM Töma Gavrichenkov <ximaera@gmail.com> wrote:

If this endpoint doesn't connect to anything outside of their network, then yes.

If it does though, the design of the filter might become more complicated.

Not really... just requires sorting by volume.  Turns out most legitimate hosts don't send high-volume syn packets. ;)

This is a good *detection* technique, but you cannot filter by volume in transit if the set of destinations is large (and random) enough, and you don't have a time machine.  Not sure if this is the case but might as well be.

They don't need to filter by destination.  Once a problem customer has been identified, they can apply an ACL restricting them to only originate IPs they own.  This was all covered in my talk at NANOG last year: https://pc.nanog.org/static/published/meetings//NANOG76/daily/day_2.html#talk_1976

As for the detection of the real source, everything is technically possible but you need certain bargaining power which a medium-sized (at best) VPN service probably doesn't have.

True, but there are ways around that, including public shaming (here), or involving law enforcement.

Damian