Re: [Sip] RE: Delivering request-URI and parameters to UAS via proxy -new version of the draft-holmberg-sip-target-uri-delivery draft

[Jonathan Rosenberg <[e-mail]>]

Sorry for jumping in late here. Trying to clarify some points here and 
there.

The UA-loose mechanism needs to know that the entity which would have 
previously been reached by the DNS/IP forwarding based on the R-URI, is 
now prepared for a request which reaches it with that target in the 
Route header field.

In the case of a UA registering through a chain of Path-inserting 
proxies, it means that only the *UA* has to support UA-loose route. 
Thats because of the assumption that the intermediate path-inserting 
proxies weren't being targeted based on the R-URI previously; they were 
(and still will) forward the request based on Route.

The 'next hop' use case that Christer is referring to are cases where we 
would have done a R-URI translation without a registration. For example, 
lets say a proxy receives an INVITE with sip:[e-mail], and its 
using some number-prefix based routing rules, which have been CONFIGURED 
into the proxy, to determine a next-hop gateway. The UA loose route 
rules would say that, since this is a routing operation and not a 
retargeting, the request looks like:

INVITE sip:[e-mail]
Route: sip:ip-of-gateway

whereas current rules would require:

INVITE sip:12345@ip-of-gateway

in this use case, the entity which was previously being reached by the 
R-URI *is* the next hop.

And so, the main limitation that I see of the UA loose route mechanism 
is, in cases where the next-hop is being reached via a configured 
routing rule, or by some other non-registration means, and said 
operation is a re-route and not retarget, the UA_loose-route requires 
that this configured routing rule include information about whether the 
next hop uses UA loose routing, and if so, what URI to use in the Route 
header.

This is a limitation, I agree. Whether its significant or not is debatable.

I will note that, it is possible to separate these concerns. We could 
define UA loose route to strictly be used for the registration cases, 
since that is the specific problem at hand.

I agree CHrister's mechanism does not have this configuration 
limitation, and that this aspect of it is better. My main beef, as 
others have commented, is that it leaves us with nearly half-dozen 
header fields which are all awfully similar (To, P-C-ID, Route, 
History-Info, and now Target). I was hoping to reduce the set and 
include only the minimum. In that way, I think UA loose route is 
architecturally better since it gives some clarity to this.

-Jonathan R.


Elwell, John wrote:
> Christer,
> 
> I don't think the entity using the mechanism needs to know about next
> hop support for the loose route mechanism - only whether the UAS
> supports it. If the UA registers, then the loose route draft provides an
> automatic mechanism. If the UA does not register, support would need to
> be indicated by provisioning. I am not sure this ranks as a significant
> limitation, since a certain amount of provisioning needs to be done
> anyway for UAs such as gateways that do not register.
> 
> John 
> 
>> -----Original Message-----
>> From: Christer Holmberg [mailto:[e-mail]] 
>> Sent: 16 January 2008 11:51
>> To: Elwell, John; [e-mail]
>> Cc: Paul Kyzivat; DRAGE,Keith (Keith); Francois Audet
>> Subject: RE: [Sip] RE: Delivering request-URI and parameters 
>> to UAS via proxy -new version of the 
>> draft-holmberg-sip-target-uri-delivery draft
>>
>>
>> Hi, 
>>
>>> I hear what you are saying, but I don't really see that the 
>> points you
>> raised in section 4 are real limitations of the loose route mechanism.
>>
>> I think it is a limitation that an entity using the mechanism has to
>> know whether the the next hop supports it or not, and that 
>> the mechanism
>> can only be used if the subsequent hops support it.
>>
>> Regards,
>>
>> Christer
>>
>>
>>
>>
>>>> -----Original Message-----
>>>> From: Christer Holmberg [mailto:[e-mail]]
>>>> Sent: 16 January 2008 11:06
>>>> To: Elwell, John; [e-mail]
>>>> Cc: Paul Kyzivat; DRAGE, Keith (Keith); Francois Audet
>>>> Subject: [Sip] RE: Delivering request-URI and parameters 
>> to UAS via 
>>>> proxy - new version of the draft-holmberg-sip-target-uri-delivery 
>>>> draft
>>>>
>>>>
>>>> Hi John,
>>>>
>>>>> Thanks for this revision, which makes things somewhat 
>>> clearer. I do 
>>>>> have a couple of comments:
>>>>>
>>>>> 1. I am not sure I agree with the assertions in section 4 
>>> concerning 
>>>>> issues with the mechanisms in loose-route. Taking example 1, the 
>>>>> Route header field should contain enough entries to get 
>> you to the 
>>>>> registered contact, not just to an intermediate proxy. 
>>> Therefore this 
>>>>> situation should not arise with a correctly implemented 
>>> home proxy. 
>>>>> It is not clear to me how example 2 could arise either, 
>>> for similar 
>>>>> reasons. The MGC case can be resolved by taking into account the 
>>>>> option tag in the REGISTER request, or if it is permanently 
>>>>> registered, through provisioning.
>>>> The examples are not meant to show bugs in the loose-route 
>>> mechanism.
>>>> They are meant to help people understand the limitations with the 
>>>> loose-route mechanism.
>>>>  
>>>>> 2. Comparing the mechanism proposed with the loose-route 
>>> mechanism, 
>>>>> my understanding is:
>>>>> a)When retargeting occurs, the loose-route mechanism 
>>> places the new 
>>>>> target in the Request URI. Your proposal places the new 
>> target in 
>>>>> both the Request-URI and the Target header field.
>>>>> b) When rerouting, the loose-route mechanism places the 
>> new route 
>>>>> (i.e., the registered contact) in the Route header field. Your 
>>>>> proposal places the new route in the Request-URI (the 
>>> latter as per 
>>>>> RFC 3261).
>>>>> So the two mechanisms solve exactly the same problems using a 
>>>>> slightly different mechanism. Correct?
>>>> Yes. The two solutions intend to solve the same problem.
>>>>  
>>>>> 3. How P-Called-Party-ID fits into this is not really 
>>> relevant from 
>>>>> an IETF perspective - it seems there are some 3GPP-specific 
>>>>> situations where the contents of P-Called-Party-ID will 
>>> not equal the 
>>>>> contents of Target. Correct?
>>>> Correct.
>>>>
>>>>> 4. If my suggestion in point 1 above that the loose-route 
>>> mechanism 
>>>>> does not suffer from the problems suggested, then each 
>>> mechanism will 
>>>>> work and each addresses the same problem.
>>>>> So it is just down to a beauty contest between the two. Correct?
>>>> See question 1.
>>>>
>>>> We believe that our solution does not have the same 
>>> limitations as the 
>>>> loose-route solution. But, again, both solutions intend to 
>>> solve the 
>>>> same problem.
>>>>
>>>> Regards,
>>>>
>>>> Christer
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>>> -----Original Message-----
>>>>>> From: Christer Holmberg 
>> [mailto:[e-mail]]
>>>>>> Sent: 16 January 2008 08:41
>>>>>> To: [e-mail]
>>>>>> Cc: DRAGE, Keith (Keith); Paul Kyzivat; Elwell, John; 
>>> Jeroen van 
>>>>>> Bemmel; Francois Audet
>>>>>> Subject: Delivering request-URI and parameters to UAS via
>>>>> proxy - new
>>>>>> version of the draft-holmberg-sip-target-uri-delivery draft
>>>>>>
>>>>>>
>>>>>> Hi,
>>>>>>
>>>>>> We've uploaded a new version (-01) of the Target draft.
>>>>>>
>>>>>> We've tried to make things more clear. I've also 
>>> removed all text 
>>>>>> about P-Called-Party-ID, except from one chapter where 
>>> we try to 
>>>>>> explain the semantical difference between Target and P-CPI.
>>>>>>
>>>>>> You can also find the draft from:
>>>>>> http://users.piuha.net/cholmber/drafts/draft-holmberg-sip-targ
>>>>>> et-uri-del
>>>>>> ivery-01.txt
>>>>>>
>>>>>> Regards,
>>>>>>
>>>>>> Christer
>>>>>>
>>>>
>>>> _______________________________________________
>>>> Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
>>>> This list is for NEW development of the core SIP Protocol Use 
>>>> [e-mail] for questions on current sip Use 
>>>> [e-mail] 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 
>>> [e-mail] for questions on current sip 
>>> Use [e-mail] 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 [e-mail] for questions on current sip
> Use [e-mail] for new developments on the application of sip
> 

-- 
Jonathan D. Rosenberg, Ph.D.                   499 Thornall St.
Cisco Fellow                                   Edison, NJ 08837
Cisco, Voice Technology Group
[e-mail]
http://www.jdrosen.net                         PHONE: [phone]
http://www.cisco.com


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