RE: .COM Clusters are Not RSCs

[Jim Fleming <JimFleming at doorstep.unety.net>]

On Thursday, March 19, 1998 4:10 PM, Steve Bellovin[SMTP:smb at research.att.com] wrote:
<snip>

@It would no longer be possible for me to advertise my mail server
@as being research.att.com.  Rather, I'd have to specify *which*
@.com domain my site was listed with -- or rather, which set of
@.com domains, since (as was pointed out above by a *proponent*
@of this scheme) there can, will, and -- in this scheme -- *should*
@be different ones.
@

This is absolutely not true. I suggest that you play with some RSCs
and .COM clusters and watch the messages being exchanged and
you will see this is not the case.

For the technical people here...:-)...here is what would happen
in Australia (as an example).

1. A stub resolver (in a PC) has a query, such as research.att.com.
2. It sends that to the ISP name server. Let's say this is the first
	name in .COM that server has seen.
3. The ISP Name Server queries the RSC it is using for the location
	of .COM. It gets the IP addresses of the .COM cluster.
4. The ISP Name Server then sends the query to the .COM cluster
	and it is told where the ATT.COM name servers are which happen
	to be in the U.S.
5. Then the ISP Name Server sends a qeury to the U.S. and the
	rest of the process is the same.

The difference in this scenario and the current scenario is that
Steps 1 to 4 happen inside Australia. This cuts down on traffic
to and from the U.S.

Now, I understand that AT&T makes money carrying that extra
traffic. This should not prevent you from studying the technology.

-
Jim Fleming
Unir Corporation
IBC, Tortola, BVI