Hadriel Kaplan wrote:
The subtle difference here -- unlike in HTTP -- is that there are
requests coming towards the client from the server.  In HTTP, the
[...]
Right, but that's why I say it's the same thing as TCP persistent
connections in this draft.  You're not allowing upstream requests
over TCP, only over TLS.  You're saying keep the TCP connection from
the client to the server open, and send subsequent requests over it
in the downstream direction (client->server).  Requests going the
other direction are sent over a TCP client->server connection from
the other side.  In that way it's exactly like HTTP/1.1.

OK; we are in vehement agreement, then.

Sure, but it doesn't solve it.  Persistent connections were always
possible, so even if this new draft says to do it by default, the
server has no way to know it is or isn't being done, and the client
has no way to be told to do it. (not that I'm advocating that be done
mind you)

It is true what you write above; i.e., in SIP, the server has no way
of knowing whether the client is using persistent connections or not,
and the client has no way of signaling its intent (unlike in HTTP,
where the signaling makes this clear.)

However, as an implementor, there is some ambiguity in rfc3261 about
how long to keep connections open.  S18 is where this information
is concentrated, and it basically recommends implementations keep
connections open for at least (T1*64)s.  And I have observed wide
variability around this: some implementations close connections
after (T1*64)s, some keep them open for longer, while others even
open up a new TCP connection to send a response (and then promptly
close it.)  So, one aspect connect-reuse was trying to establish
was to establish a default behavior so that the connection stays
open for as long as possible; certainly beyond (64*T1)s, and in
fact, as long as the resources on a machine allow that connection
to remain open.

- vijay
--
Vijay K. Gurbani, Bell Laboratories, Alcatel-Lucent
2701 Lucent Lane, Rm. 9F-546, Lisle, Illinois 60532 (USA)
Email: [EMAIL PROTECTED],bell-labs.com,acm.org}
WWW:   http://www.alcatel-lucent.com/bell-labs


_______________________________________________
Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [EMAIL PROTECTED] for questions on current sip
Use [EMAIL PROTECTED] for new developments on the application of sip

Reply via email to