> UCBCMSA --  UCCVMA  --  site 1
>  (core)     (us)   \
>                     --  site 2 (and another via a separate link)
 
> We intend to leave Bitnet and license Listserv TCP/IP and LMail.  The
> downstream sites may or may not leave Bitnet.  All have LMail and VM
> TCP/IP, though one site is not running the VM SMTP.
 
If the downstream sites intend to stay in BITNET, they'll need to
connect to another site upstream from you. The sites running
VMNET won't have too much trouble with this, but the SNA sites
will be more difficult -- but you probably already knew
that...8-).
 
Since they already have IP connectivity (modulo the site that
doesn't run SMTP), you should encourage them to configure their
mailer to emit only FQDNs ASAP. This will reduce the amount of
mail sent via NJE (at the expense of SMTP cycles and I/O
overhead). For the site that doesn't run SMTP, you can configure
your mail system to be the primary MX for that site and then
deliver the mail to them via your private NJE hookup by some
judicious hacking our of the OVERRIDE files.
 
>  What can I do, if anything, to minimize the
> impact on the downstream sites?
 
For NJE SENDFILE, there's not much you can do. For mail, get them
to start emitting FQDNs as soon as possible. If their mail agent
doesn't do FQDNs, get them started on a migration to something
that does Real Soon Now (like Mailbook or something like that).
 
>  Do they need to connect to an NJE site
> upstream from me to minimize disruption?
 
If they don't want to change anything, or they want SENDFILE to
continue to work, then yes, they need another NJE connection.
 
Drop me a note privately if you want to discuss this in more
detail.
 
-- db