Re: Is round-trip time no longer a concern?

Keith Moore <moore@cs.utk.edu> writes:

>> Well, I'm not claiming that latency isn't a factor in protocol
>> performance. What I'm claiming is that it's not clear that latency
>> in the initial connection setup handshake (in this case the TLS
>> one) is a major factor in protocol performance. Recall that the
>> way that TLS works is that you do an initial handshake to establish
>> the keys and then you send data of the negotiated channel. What
>> we're discussing is whether it's OK for this to take a few extra
>> round-trips at the setup of the first connection (and not necessarily
>> afterwards because of TLS session caching). So, we're not talking
>> about adding messages to every activity of the higher level protocol.
>
> True, but this presumes a lot about how TLS is used.  For instance, if
> you're using it with HTTP to load a web page that downloads not only a
> root HTML file but lots of other files for frames, images, applets,
> etc., then you're going to incur that connection setup overhead for
> every one of those objects.  And because (partially for this reason)
> with HTTP it's fairly normal for a browser to open multiple
> connections to a server site, all of those delays won't be serialized,
> but they'll still impact the total time to download the page.
>
> Now maybe this specific example illustrates a flaw in the design of
> HTTP and/or HTML, but the extra setup overhead of TLS (and this new
> extension) does impact its versatility.
>
> Also I chose HTTP/HTML because most people are familiar with it.
> There are other, more obscure, applications involving lots of hosts
> that need to open up secure and sometimes brief connections with one
> another.
>
> So while the extra setup delays of TLS might not matter much in some
> applications, they definitely matter in others.

TLS incorporates a "session resumption" feature that is specifically
designed to let you avoid a new handshake for each connection you
make (that's what the "not necessarily afterwards" above meant.)
So, the full TLS cost is only paid for the first connection to
a given server. 

-Ekr

_______________________________________________
Ietf mailing list
Ietf@ietf.org
https://www1.ietf.org/mailman/listinfo/ietf