The current draft, draft-ietf-sip-outbound-00.txt, proposes to use STUN for
UDP and CR/LF for TCP. However, the I believe the next version of the draft
will recommend STUN for both UDP and TCP. At IETF 63, it was noted that the
CR/LF mechanism will not work because it may take several minutes for the
client to realize the TCP connection has failed.

cheers,
(-:bob

Robert F. Penfield
Chief Software Architect
Acme Packet, Inc.
71 Third Avenue
Burlington, MA 01803
[EMAIL PROTECTED]


----- Original Message ----- 
From: "Paulo de Arruda Borelli" <[EMAIL PROTECTED]>
To: "'Hyoungjoon Park'" <[EMAIL PROTECTED]>
Cc: <[email protected]>
Sent: Tuesday, August 16, 2005 7:28 AM
Subject: RE: [Sip-implementors] keep alive with NAT in SIP


> Hi, Jun,
>
> Are you aware of any recommendation for SIP over UDP?
>
> -----Original Message-----
> From: Hyoungjoon Park [mailto:[EMAIL PROTECTED]
> Sent: Tuesday, August 16, 2005 7:11 AM
> To: [EMAIL PROTECTED]
> Cc: [email protected]
> Subject: RE: [Sip-implementors] keep alive with NAT in SIP
>
> Hi, Paulo,
>
> I think IETF has recommended the CR/LF mechanism for SIP over TCP,
although
> I'm quite not sure for NAT case.
> Please check out the following link;
>
>  http://www.softarmor.com/sipwg/meets/ietf61/notes/minutes-sip-ietf61.html
> "1. TCP keepalive every periodic number of seconds? CRLF?
>  REGISTER? New message (PING)?
>       Comments:
>  - REGISTER is horrifically expensive - not a general solution
>  - CRLF doesn't generate an application ACK - have to wait for TCP
>  keepalives. question about if standard TCP interface allows check.  We
>  think we can check unacked bytes in most(?) socket interfaces and use
CRLF.
> "
>
> Regards,
> Jun
>
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] On Behalf Of Paulo de
> Arruda Borelli
> Sent: Saturday, August 13, 2005 4:00 AM
> To: [email protected]
> Subject: RE: [Sip-implementors] keep alive with NAT in SIP
>
> Hi, Rakesh,
>
> You can use the expires value of the REGISTER messages. Just configure the
> proxy to require a low expires value (like 60 seconds). This will force
the
> endpoint o re-register again every 60 seconds. In fact, in less than 60
> seconds - because, if they re-register exactly every 60 seconds, they will
> certainly lose registration until the new registration gets active. Some
> endpoints will re-register at 50% of expires (30 seconds, in this case);
> some will at 90%. But they usually re-register at less than the nominal
> expires value.
>
> The idea is that this will force the NAT pinhole to be open all the time -
> enabling inbound flow (from the proxy to the endpoint) to reach the
> endpoint.
>
> Some brands (of endpoints) also support the OPTIONS method as keep-alive
> mechanism. This will also keep the NAT pinhole open. But the re-register
> method is great - since it's usually supported. Notice it's imperative
that
> the proxy force expires to a low value. Usually, the endpoints send a
> default value of 1800 or 3600 seconds - and not always configurable. To
keep
> the NAT pinhole open, it's imperative to re-register at less than one
> minute.
>
> Paulo Borelli
>
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] On Behalf Of Nataraju A
B
> Sent: Friday, August 12, 2005 5:37 AM
> To: 'Rakesh Dhandhukiya'; [email protected]
> Subject: RE: [Sip-implementors] keep alive with NAT in SIP
>
>
>
> Regards,
> Nataraju A.B.
>
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] On Behalf Of Rakesh
> Dhandhukiya
> Sent: Tuesday, August 09, 2005 1:03 PM
> To: [email protected]
> Subject: [Sip-implementors] keep alive with NAT in SIP
>
> Hi !
>
> I want to keep alive with NAT. First of how to know the keep alive timeout
> of NAT for UDP. Is there any file like tcp timeout ?
> Is ICE used for NAT keep alive?
>
> [ABN] service provider may not disclose these timer values... only way to
> find out these timer values would be to use trail and error method....
> I don't think ICE been used to NAT keep alive...
>
> Thanks.
>
> Rakesh Dhandhukiya
> _______________________________________________
> Sip-implementors mailing list
> [email protected]
> http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
>
> _______________________________________________
> Sip-implementors mailing list
> [email protected]
> http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
>
>
>
> _______________________________________________
> Sip-implementors mailing list
> [email protected]
> http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
>
>
>
>
>
> _______________________________________________
> Sip-implementors mailing list
> [email protected]
> http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
>

_______________________________________________
Sip-implementors mailing list
[email protected]
http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors

Reply via email to