Re: [quicwg/base-drafts] Pad path validation in both directions (#4241)

[Christian Huitema <[e-mail]>]

The proposed PR bakes the assumption that paths should be symmetric. This might be an OK restriction for V1, but I hate seeing that baked in the code. There are several scenarios in which data can be send one way on a broadcast channel (e.g., satellite or DVB), while the other way is through a slower but symmetric channel.

Supporting asymmetry requires testing each direction. The server, instead of changing the address at which it sends merely because it received data from the client on that new path, should perform its own Challenge/Response to validate the path. Since challenges have to be padded, this tests both continuity and MTU. The server learns that a path may exist by receiving packets on it; it only uses that path for sending if the path is validated.

I guess this is something we will need to revisit as part of support for multipath.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/pull/4241#issuecomment-713086130