Bob and Orit,
Indeed, it would appear that this "proto" or "mode" parameter is still a bit contentious.
Let me make a few comments.. my immediate goal is to see H.323v4 brought to Decision. We can haggle over Annex O (which will contain all of the specific procedural text) later on. I agree with Bob-- we will have a few of these meetings, I am sure.
There is nothing in the URL scheme itself that would prevent a Gatekeeper or an endpoint from utilizing DNS. When the mode is specified as "direct", we specified that the hostname resolves has an associated 'A' record in DNS. However, if I give you h323:paulej@packetizer.com, you can't assume that 'packetizer.com' will have an 'A' record-- and if it does, it may not contain H.323 entity (endpoint, GK, or BE). So, in the "indirect" mode (the default), the endpoint must somehow turn this URL into a callable address. Now, the procedure for this may involve DNS or it may not. The text I last sent out suggested that this is the "domain name", but I did not say DNS or "administrative domain name". I can imagine that either may be utilized, but I also prefer that the "domain name" refer to a DNS domain name and would also like to propose that the "domain name" mentioned in this section be qualified with "DNS domain name": that might not sit well-- Bob?
How about this: what if I remove the text that says that "indirect" means that "an endpoint shall use a Gatekeeper to resolve the address". Instead, I will say "the mode value indirect indicates that the address does not refer to a callable H.323 entity. The procedure for resolving such an address is for further study." Making this change does not prevent a Gatekeeper from being involved and it also does not prevent an endpoint from doing some type of direct resolution. Perhaps this will get us past this-- we'll debate this when we get to Annex O.
Orit, one thing you say over and over again is "direct" and "routed". I never used the word "routed"-- do not mistake "indirect" for "routed". The word "direct" versus "indirect" merely indicated that the entity is callable or not.
Perhaps this: in addition to making the above change, we change "mode" to "callable". The URL will then take this form:
h323:paulej@packetizer.com;callable=no or h323:paulej@paris.packetizer.com;callable=yes
The default would be "callable=no". This equates to the same thing as "mode=direct" and "mode=indirect", but perhaps it's clearer? (Or perhaps not... I personally prefer "mode=[in]direct").
Paul
----- Original Message ----- From: Callaghan, Robert To: 'Orit Levin' ; Paul E. Jones Cc: Mailing list for parties associated with ITU-T Study Group 16 (E-mail) Sent: Friday, September 08, 2000 4:45 PM Subject: RE: Re: Revised H.323 URL document
Orit,
Based on this email, we have not agreement at all.
1) I am not interest at all in developing an addressing scheme for the sole reason that it mimics the addressing schemes used by SIP. SIP is very oriented towards the concept of domains defined by DNS and a user in that domain. H.323 is oriented towards administration domains defined by one or more gatekeepers and users attached to those gatekeepers. The only commonality is that a gatekeeper can be a server with a name defined by DNS. These distinctions must remain.
2) It is unacceptable to force the implementation of annex G on any endpoint, gatekeeper or administrative domain. Therefore the differentiation between the use of Annex G or the use of LRQ should not known to the originating endpoint. The differentiation should be that this address defines the destination endpoint (direct relationship) or it defines a source for resolving the destination endpoint address (indirect).
In my view, in the direct relation a SETUP message is routed towards the entity defined by the DNS where the user portion is resolved. This may be direct routed or gatekeeper routed, it does not matter. The endpoint defined by the DNS may be a gatekeeper or the endpoint itself where the user portion defines one of multiple users on that endpoint.
For the indirect relationship, the endpoint must resolve the address of the destination address before sending the SETUP message. Based on H.323 procedures, the endpoint sends ARQ to its GK. How the GK resolves the address is within the capabilities of the GK. It may use Annex G, it this is possible. Or, it may fall back to LRQ to the entity specified by the DNS name.
3) The definition of an H.323 URL cannot be the vehicle to implicitly define new procedures for address resolution. If a new procedure for address resolution is to be defined, it must be explicit, and it must be clear. These requirements have not been realized. I have seen no procedural text that has full agreement. I have seen Annex O as a working document, but this is a year from closure. I expect to have many discussions on this subject in future meetings (note the plural). The H.323 URL we have for November must support the existing H.323 procedures without any implicit extension or modification.
I think that Paul's proposal meets the requirements that I have stated.
Where does this leave us now?
Bob
------------------------------------------------------------------
Robert Callaghan
Siemens Enterprise Networks
Tel: +1.561.923.1756 Fax: +1.561.923.1403
Email: Robert.Callaghan@ICN.Siemens.com
------------------------------------------------------------------
-----Original Message----- From: Orit Levin [mailto:orit@radvision.com] Sent: Friday, September 08, 2000 3:57 PM To: Paul E. Jones; Robert (Bob) Callaghan Subject: Re: Re: Revised H.323 URL document
Hello Paul and Bob!
I prepared the following "speech" for the mailing list, but it is definitely better to try to understand each other first...
---------------------------------------
I hate to disappoint you, but I would strongly disagree with the proposed "mode" terminology.
I will try to explain below, why it does not fit the URL concept.
1. The MAIN purpose of URL is to use the Internet DNS Infrastructure for SERVICE+ADDRESS resolution ACROSS the DOMAINS.
(The use of the "user-only" URL is [just] a possible additional feature.)
2. The H.323 "direct vs. routed" concept specifies the relations between an End Point and ITS Gatekeeper. It is an local policy issue WITHIN their zone.
Today, according to H.323 specification, the real flavor of the operational mode in the DESTINATION zone is UNKNOWN (and irrelevant) to the originator of the call.
Not mentioning the future possible GK decomposition!...
3. Following the proposed "mode" terminology, I guess, "routed" would mean "LRQ" or "Access Request" and "direct" would mean "Setup". In reality, the LRQ would lead to the "direct call" and the Setup would lead to the "routed call". And how to distinguish between the all three?
4. The INTER-ZONE communication haven't been solidly defined so far. (LRQ ? Annex G?)
The INTER-DOMAIN communication is the subject of H.225.0 Annex G.
The URL approach is an OPTION for all of the "intra" and "inter" H.323 Zone and H.323 Domain cases.
It may be used as a first lookup, before the LRQ or Annex G are performed.
It may also have an overlapping functionality with LRQ and with very specific PARTS of Annex G.
The relations between the H.323 URL and H.225.0 "Annex G" are very close to those of the SIP-URL and TRIP in a "SIP system". Both have their uses and complement each other.
5. The proposal is to have a possibility to "publish", using the URL, the first procedure, to be issued towards the destination "domain".
By its nature, it is not an issue for standardization. It is a policy of the destination domain, which also is responsible for the URL itself.
The first procedure (i.e. the "method") required for establishing of a specific call across domains/zones can be either of the following:
- H.225.0 Annex G/AccessRequest
- H.225.0/LRQ
- H.225.0/CALL SETUP
Once, the first message is issued towards the destination (which is a result of DNS lookup using the URL), the rest of the procedures are, indeed, very well defined by H.323 and H.225.0 Annex G all together.
Best Regards,
Orit Levin RADVision Inc. 575 Corporate Drive Suite 420 Mahwah, NJ 07430 Tel: 1 201 529 4300 (230) Fax: 1 201 529 3516 www.radvision.com orit@radvision.com
-----Original Message----- From: Callaghan, Robert Robert.Callaghan@ICN.SIEMENS.COM To: ITU-SG16@MAILBAG.INTEL.COM ITU-SG16@MAILBAG.INTEL.COM Date: Thursday, September 07, 2000 8:17 AM Subject: Re: Revised H.323 URL document
Paul,
I definitely like "mode" as a replacement for "protocol." It clearly represents the required procedure without over doing details.
I think that we got it.
Bob
------------------------------------------------------------------
Robert Callaghan
Siemens Enterprise Networks
Tel: +1.561.923.1756 Fax: +1.561.923.1403
Email: Robert.Callaghan@ICN.Siemens.com
------------------------------------------------------------------
-----Original Message----- From: Paul E. Jones [mailto:paulej@acm.org] Sent: Wednesday, September 06, 2000 11:20 PM To: Robert Callaghan; Orit Levin; ITU-SG16@mailbag.cps.intel.com Subject: Revised H.323 URL document
Orit, Bob, et al,
Attached is the revised section covering the syntax of the H.323 URL. This (I hope) includes all of the changes we have discussed so far. I am not sure that we settled on the "mode" parameter, but I made the change here in this document-- feedback is welcome.
Best Regards,
Paul