Eric,

Your draft in fact addressed the issue that I'm concerned about.  I'm focused 
on using something like TCPINC in kernel-to-kernel TCP communication.  Your 
section 5 calls out implementation of the kernel option as "more effort for 
kernel developers".  In my case, there's no user-space application to which the 
kernel can kick the negotiation of TLS, so the kernel option is the only option.

Craig

From: Eric Rescorla <[email protected]<mailto:[email protected]>>
Date: Tuesday, July 29, 2014 11:19 AM
To: Craig Everhart <[email protected]<mailto:[email protected]>>
Cc: Christian Huitema <[email protected]<mailto:[email protected]>>, Mark 
Handley <[email protected]<mailto:[email protected]>>, 
"[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>
Subject: Re: [tcpinc] Protect or not the TCP header




On Tue, Jul 29, 2014 at 7:58 AM, Everhart, Craig 
<[email protected]<mailto:[email protected]>> wrote:
On 7/29/14 1:58 AM, "Christian Huitema" 
<[email protected]<mailto:[email protected]>> wrote:

>Let look at TCP + crypto. It has to compete with two established
>standards.
>On one hand, TLS, which is easily deployed but does not protect any of the
>TCP headers. On the other hand, IPSEC, which is harder to deploy but does
>protect the TCP header. A secure version of TCP only makes sense if it
>protects the headers better than TLS and is easier to deploy than IPSEC.

I don't think TLS is so easy to deploy.  TCP is used in a wide variety of
environments; that's one of its attractions.  One of the attractions of
TCPcrypt is its aim to be deployable wherever TCP can be, which is a bigger
space than where TLS is convenient.  That's my hope for the WG.

Can you say more about why TLS is hard to deploy? I'm particularly interested
in issues which aren't addressed by my draft.

Thanks,
-Ekr

_______________________________________________
Tcpinc mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tcpinc

Reply via email to