Re: [TLS] Request for review: Next Protocol Negotiation Extension

[Marsh Ray <[e-mail]>]

On 08/18/2010 03:26 PM, Michael D'Errico wrote:
> Adam Langley wrote:
>> However, the number of round trips is critical. However fast the hosts
>> might be, the speed of light always stays the same.
>
> So then NPN is all about reducing a round trip? You say you already
> achieve that with false start. Or do you need something in addition
> that can only be achieved by sticking it in a TLS extension?

If you're using false start, then the inital application data can go out 
in the same packet as a handshake message before Finished, at least in 
one direction, right?

If I understand correctly, the cross protocol DoS prevention is 
accomplished by requiring the client to send something to the server to 
unambiguously signal the use of a specific protocol.

Can't you use some nonces already present in the handshake without 
extending the protocol? A new handshake message is a big deal, I don't 
think it's been done since SSLv2->v3.

For example, couldn't you have the client and server exchange something 
like:

"[protocol: WebSockets " + MD5(TLS session ID  + "WebSockets") + "]\r\n"

as the initial false start application data?

> THAT is what you should be working on. Prediction. If I've clicked
> the same link every morning for the past 18 months, what are the odds
> that I'm going to click it today? 100%. And even if for some reason
> I don't, I'm not going to be at all upset that my browser fetched the
> page for me.

Sounds like you were reading:

http://online.wsj.com/article/SB10001424052748704901104575423294099527212.html

No thank you :-)

- Marsh