I'm a Principal Engineer in the Office of the CTO at Fastly. Formerly R&D at the RIPE NCC, IPv6 deployment at Yahoo, network measurement at Boundary. PhD from University of Glasgow in 2012.
View My GitHub Profile
Public activity
Archive
IPv6
- continuing the IPv6 target generation work
- IPv6 header comparison
- triggered by another IPv4-only paper
- record route doesn’t really exist; what have people written drafts for or prototyped?
- critically: what actually passes? Does the routing header actually pass through? Can it be used?
- related header extension work at TMA2017
- discussed briefly with authors of this IMC2017 paper
- open firewall policies: what actually passes through?
- can, say, Atlas probes send unsolicited pings to each other (ICMP/UDP/TCP?)? Addressing may be global, but what firewalling exists, and where?
- related: the PMTUD problem
- the KSK rollover
- IPv6 diffs
- diff between originating ASNs in BGP dumps
- diff in AS hops between pairs of hosts
- diff in AS paths between pairs of hosts: peering distinctions
- IPv6 and anycast stability
- Wei and Heidemann: anycast to DNS roots on v4
- but the v6 question is left unanswered, and it’d be really interesting
- I imagine it’s the same in most cases, but peering will differ and that may affect behaviour (for better or worse)
- Spoofing:
- this paper is great but doesn’t touch on the v6 data
- Threats in the IPv6 world
- botnet traffic, amplification attacks, continuous sweeping: all commonplace in IPv4
- the IPv6 space modifies how people attack it or measure it
- so, what does this traffic look like in the IPv6 side? does it exist at all?
- v6 growth using atlas probes
- survey scanning techniques
IPv4
- better studies of IPv4 transfer markets
Archaeology
- Things get deprecated in standards. But what still works?
- conventional wisdom was that IPv4 Record Route was useless, but a paper in 2017 refutes that: the Record Route Option is an Option
- So what else works?