[tor-dev] How does Tor plan to deal with HTTP/3 (HTTP over QUIC)

Iain Learmonth irl at torproject.org
Sat Nov 17 16:40:20 UTC 2018


On 15/11/18 02:02, neel at neelc.org wrote:
> How would Tor deal with HTTP/3 (a.k.a. HTTP over QUIC), considering that Tor is a TCP anonymizer, and HTTP over QUIC (and QUIC itseld) uses UDP? Would we need Tor to support UDP? Just QUIC?

One reason we don't support UDP in Tor because it is connection-less,
and the connection concept in TCP means that the server at the other end
needs to accept the connection before we start sending larger amounts of
data to it.

Saying this, QUIC is not UDP. It may use UDP for the sake of middlebox
traversal but is in fact a connection-oriented transport protocol. (If
you ignore the unreliable datagrams draft currently in the IETF QUIC WG).

This recent presentation at the IETF looks at some of the ways that QUIC
and HTTP/QUIC can be tunneled from the perspective of the client
interface and the exit relay:


To really get benefits from things like streams, it could be necessary
to map some of the QUIC internals into the Tor protocol.

Note that SOCKS 5 actually does already support UDP, we just don't
implement this part of the protocol for clients.

I'm not a Network Team person so I don't know if this is currently
something being considered, but given that the IETF hasn't actually
standardized QUIC yet this is probably not going to be needed urgently.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://lists.torproject.org/pipermail/tor-dev/attachments/20181117/1e54bd0c/attachment.sig>

More information about the tor-dev mailing list