[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: Tor speed



Hi!

On Fri, Feb 13, 2009 at 12:55 PM, Scott Bennett <bennett@xxxxxxxxxx> wrote:
> By reinventing TCP within tor, the proponent(s) believe they can achieve
> better overall performance, even though they lose the basic operating-system
> level of support for reliable delivery in the process and I think they lose one layer
> of encryption, as well.

I would go with this question further: why reinventing TCP over UDP
again in general? General solutions and implementations for that
already exist. For example, Tor could just simply add another layer
with OpenVPN (or just OpenVPN based) system.

One hackish, to show my idea, solution could be to build first a world
wide VPN network based on OpenVPN with internal IPs, each Tor node
would get one. Modify Tor nodes to open a new TCP circuit (even if it
already exists) with a target node for every data stream. And this is
it. Data streams would not be multiplexed in one TCP session (hm, does
this decrease security?) so if one TCP stream has packet loss this
will not influence other, and rebuilding a TCP connection from UDP
packets would be done by OpenVPN.

I am not suggesting to implement this. I am just saying that a lot of
(open source) work of making TCP over UDP has already been made.

> They may be correct about that, but I am still wondering why UDP has been
> chosen over SCTP, which looks to me as though it would provide just about
> anything alternative that might be useful in improving throughput, reliable
> delivery, and signalling, but that is a question that no one has responded to
> so far.

Maybe because SCTP is not yet widely supported?

Good reading about TCP over TCP:

http://sites.inka.de/~W1011/devel/tcp-tcp.html


Mitar