Turbo Tunnel Q&A and workshop

Internet Archive video mirror

In short: circumvention protocols should include an internal session and reliability layer, one that is decoupled from the obfuscated carrier session. There are many benefits to doing so, but the main advantage is design flexibility that enables us to write more adventurous and complexly layered protocols.

Background:

• https://www.bamsoftware.com/papers/turbotunnel/ short paper and video from 2020
• https://www.bamsoftware.com/papers/turbotunnel/example/ worked example of a Turbo Tunnel conversion with extensive comments

Demo for today: quick Turbo Tunnel conversion

• https://www.bamsoftware.com/talks/ptim-2021-turbotunnel/ptim-2021-turbotunnel-demo.zip

There are many options for an internal session protocol (https://github.com/net4people/bbs/issues/14). The demo today uses KCP+smux:

KCP

provides a reliable stream interface over unreliable datagrams

smux

connection termination after timeout, stream multiplexing

Success stories:

• Snowflake (peer-to-peer WebRTC) - https://github.com/net4people/bbs/issues/35
• dnstt (DNS tunnel) - https://www.bamsoftware.com/software/dnstt/
• Champa (AMP cache tunnel) - https://repo.or.cz/champa.git