Dear Colleagues:

I completely agree with Mr. Roy that we should include a discussion in Annex G on how zones within an Administrative domain can be arranged as is described in APC-1385.

However, my version of APC-1375 titles Annex G as "Communication Between Administrative Domains". As currently defined in Annex G, an Administrative Domains includes a "Border Element" and there is an implied hierarchy structure between a border element and regular gatekeepers (granted that this could be a degenerate form of hierarchy, i.e. flat).

The proposal we brought to the table is how we could arrange these Administrative Domains in a manner that would be scalable consistent with Goal #2 as stated in Annex G.

Santo Wiryaman Videoserver Inc. swiryaman@videoserver.com +1.781.505.2348 +1.781.505.2101 (fax)

---------- From: Mailing list for parties associated with ITU-T Study Group 16 [SMTP:ITU-SG16@mailbag.jf.INTEL.COM] Sent: Thursday, July 23, 1998 8:50 AM To: ITU-SG16 Subject: Re: H.225 Annex G topics

Hi Everyone:

This proposal of Videoserver suggests a specific implementation scheme that is known as "hierarchical (or centralized)" model. A hierarchical model has both advantages and disadvantages (we had many discussions on this before).

However, Annex G does not impose any restriction how the system organization will look like. Consequently, the system organization can be "non-hierarchical (or distributive)", "hierarchical (or centralized)", or "hybrid (distributive + centralized)" as proposed in APC 1385, and can be chosen depending what is best from an implementation point of view.

The important point is that a "zone" having a single gatekeeper (GK) remains the basic building block. A domain may consist of multiple zones. The possible logical organizational relationship for communications between the multiple zones can be "distributive", "centralized", or "hybrid" as proposed in APC 1385. The fundamental questions are as follows:

* Can we define the inter-GK communications protocol (IGCP) that satisfies the requirements in "distributive", "centralized", and "hybrid" system organizational environment? * Can we use the same RAS protocol that will provide the scalable solution for both connection oriented and connectionless networking environment considering the large-scale H.323 network? Or do we need the extension of this RAS protocol?

Therefore, it is strongly suggested that the above points that are being suggested in APC 1385 should also be included in the work scope of Annex G.

Thanks and regards,

Radhika R. Roy AT&T Tel: (732) 949 8657 Fax: (732) 949 8569 Email: rrroy@att.com