Security trivia: GTSM sets BGP’s TTL to 255 so received packets must arrive at 255, proving the peer is one hop away and blunting off-link spoofing. In multi-hop eBGP or route-reflector designs, where do you lean on TCP MD5/AO and strict interface ACLs instead to keep the control plane tight?
In multi-hop I lean on TCP MD5/AO plus CoPP, and a per-neighbor ACL that only allows TCP/179 from the exact peer IPs (“permit tcp host A host B eq 179”) with logging on drops. If the path is fixed I pin ebgp-multihop to the exact hop count and keep sessions on loopbacks — belt-and-suspenders that shrinks spoof space; do you also rate-limit 179 and alert on MD5/AO failures?
On multi-hop where the RR path is stable, I keep the ‘TTL 255’ idea by enabling GTSM with an explicit hop count (neighbor ttl-security hops 2) so off-link spoofing still gets dropped alongside CoPP. If the hop count ever shifts during maintenance, skip it there and lean on your keys/ACLs instead. Does your RR path ever vary?