RE: [Sip] moving UDP support out of the core SIP document

["Drage, Keith (Keith)" <drage@lucent.com>]

Comments/questions as follows:

1)	Your proposal deals with the change of UDP from mandatory to optional at the UE. What changes, if any, would you expect to make to proxy support and handling?

2)	I did a quick look at the RFCs for TCP, and failed to work out whether all IETF compliant TCPs are now RFC 2914 compliant or not. This is because I am not sure whether the term "updates" used in the RFC index, automatically obsoletes the previuos version, and therefore is it possible to claim IETF compliance by implementing the core TCP RFC without implementing the congestion control extension. I assume that what we would require is mandatory support of a TCP that is RFC 2914 compliant.

3)	Given that you intend to retain support of UDP, although as an optional feature, I am not sure I would agree to moving it to a separate RFC. By all means move it and the supporting procedures to a normative annex, but moving it to a separate specification will lead to endless requests on the mailing list by people asking where the UDP support has gone.


Keith


Keith Drage
Lucent Technologies
Tel: +44 1793 776249
Email: drage@lucent.com 

> -----Original Message-----
> From: Rohan Mahy [mailto:rohan@cisco.com]
> Sent: 23 October 2002 19:02
> To: sip@ietf.org
> Subject: [Sip] moving UDP support out of the core SIP document
> 
> 
> Hello Everyone,
> 
> I wanted to share some thoughts that I had about removing UDP 
> from the 
> core SIP document when we go to draft standard. We are probably at 
> least one year away from going to Draft Standard. I hope that during 
> this time most implementations will support RFC3261, and therefore 
> support TCP.  In addition, it is quite likely that when 
> moving to Draft 
> Standard that the IESG will REQUIRE TLS support on SIP User Agents. 
> This would ensure a ubiquitous hop-by-hop channel security mechanism, 
> but which requires a reliable transport (TCP or SCTP).
> 
> Obviously UDP is still very attractive in wireless networks 
> (due to RTT 
> issues) and in other typically private networks with well-understood 
> delay and loss characteristics. To make most implementations simpler 
> however, I am proposing that we make unreliable transports (UDP and 
> DCCP) of SIP optional in User Agents, and that we put the 
> UDP-specific 
> mechanism (ex: much of the complication of the transaction layer for 
> handling retransmissions) in a separate Draft Standard 
> document.  It is 
> very likely that if we did this there would also be a 
> baseline security 
> mechanism for UDP as well (probably some profile of IPsec).  The net 
> effect is that SIP User Agents could also talk to each other 
> using TCP 
> or TLS+TCP, and that User Agents with a strong preference for 
> UDP would 
> have a consistent security mechanism.  Proxies would still 
> support both 
> transports, so you could easily communicate between a TCP-only User 
> Agent, and as User Agent which prefers UDP without requiring the User 
> Agent to switch over to TCP.
> 
> I welcome your discussion and comments on this idea.
> 
> thanks,
> -rohan
> 
> _______________________________________________
> Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
> This list is for NEW development of the core SIP Protocol
> Use sip-implementors@cs.columbia.edu for questions on current sip
> Use sipping@ietf.org for new developments on the application of sip
> 
_______________________________________________
Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use sip-implementors@cs.columbia.edu for questions on current sip
Use sipping@ietf.org for new developments on the application of sip