4 ms·

>Who is working on better routing protocols? Scion is a clean slate internet architecture research design (including routing)[0]. >It seems like bgpsec is abo

>Who is working on better routing protocols? Scion is a clean slate internet architecture research design (including routing)[0]. >It seems like bgpsec is about as good as we are going to get. I was author on a paper [1] that reduced some of the downsides to the RPKI (the PKI BGPSEC relies on). There is also interesting work on getting most of the security from BGPSEC without complete BGPSEC deployment [2]. In a different direction, as we build the internet of the inner solar system we will need protocols with different properties than those we needed for terrestrial networks. [0]: https://www.scion-architecture.net/ https://www.scion-architecture.net/ [1]: "From the Consent of the Routed: Improving the Transparency of the RPKI" http://cs-people.bu.edu/heilman/sigRPKI_full.pdf http://cs-people.bu.edu/heilman/sigRPKI_full.pdf [2]: "Jumpstarting BGP Security with Path-End Validation" http://dl.acm.org/citation.cfm?id=2934883&dl=ACM&coll=DL&CFID=773220959&CFTOKEN=80733804 http://dl.acm.org/citation.cfm?id=2934883&dl=ACM&coll=DL&CFI...

hueving 9y ago
I've seen scion before but IMO "clean slate" protocols are DOA when it comes to displacing BGP. Just look at the mailing list. A protocol with an integration model of forming an overlay has no integration model. ETHZ and Co have been marketing that at all kinds of academic conferences and it's seen approximately 0 uptake outside of organizations that volunteer to run it as an experiment. Sorry about the rant, but I get the impression that the authors of replacement protocols like these are more interested in becoming academically famous for being the inventor of the Internet (i.e. The next Vincent Cerf) rather than proposing solutions to existing systems. I would be interested in new protocols for super high latency networks (i.e. The solar system model). The RPKI work and path end validation work is interesting and I haven't seen those papers before (been disconnected from publications recently). Thanks for the links.

hueving 9y ago

I've seen scion before but IMO "clean slate" protocols are DOA when it comes to displacing BGP. Just look at the mailing list. A protocol with an integration model of forming an overlay has no integration model. ETHZ and Co have been marketing that at all kinds of academic conferences and it's seen approximately 0 uptake outside of organizations that volunteer to run it as an experiment. Sorry about the rant, but I get the impression that the authors of replacement protocols like these are more interested in becoming academically famous for being the inventor of the Internet (i.e. The next Vincent Cerf) rather than proposing solutions to existing systems. I would be interested in new protocols for super high latency networks (i.e. The solar system model). The RPKI work and path end validation work is interesting and I haven't seen those papers before (been disconnected from publications recently). Thanks for the links.