I'm feeling stupid about this, but I seem to remember that the information I am looking for is already in a MERIT or some other report, and I can't remember which it is. Disclaimer, on which I expand below: I do have commercial reasons for asking this, but I also hope to make the commercial results into something generally useful. Specifically, I am trying to determine some counts for: AS that generate advertisements only for their own AS. The assumption here is these will be either multi-homed enterprises or non-transit ISPs. This could be refined with a verification that if AS1 only advertises itself, it is actually multihoming only if the path AS1 ASx is advertised by at least 2 other AS. AS that provide transit. Ideally, I'd like to separate this into "large" and "not-large," the distinction being those that require scalable IBGP technologies beyond full mesh. Why am I doing this? Crass commercial reasons, but involving information that I am quite willing to share with NANOG, and which I believe is of benefit to the community. As many of you know, I work for an internetworking training organization. On my own, I have drafted some BGP and related courses, aimed at getting exterior routing beginners to the journeyman level. With the growth of ISPs, both expansion in existing ones and new entries to the market, some of which we may feel are clueless, I believe there is a need for education beyond that which is feasible as on-the-job. Before progressing further, I need to establish to my management that there is a training market here, and to the BGP-using community that it is possible to do meaningful training in the basics, beyond a brief lecture-only model but not to a level that builds a skilled exterior routing architect. I do hold personal copyright to the first versions of the courses. While obviously I'm looking for return on investment, as would my management, I personally believe it is of value to make such training available. I would, in fact, be hoping to find part-time instructors among the established exterior routing community. I'll be happy to share the curriculum model I have developed, and feedback is more than welcome about this whole idea. Fundamentally, it is a personal goal that may become commercial. Howard Berkowitz