On Mon, 30 Mar 1992, Phillip Sollins <[log in to unmask]> wrote:
>
> 1) It looks to me like I'm being told to move the Puerto Rico folks to ucrvm2
> (San Jose, Costa Rica).  True?  If so, why? I checked uprenet with the show
> path command and it appears to be linked to ucrvm2 via upr1, servax, and
> fauvax (Florida Atlantic).  What would be the advantage then of sending mail
> to Puerto Rico via Costa Rica?
>
The output from "SHOW PATH" is:
 
YALEVM -> UMDD2 -> NER -> SERVAX -> (9.6K) -> UPR1   -> (1M)   -> UPRENET
                                 -> (64K)  -> FAUVAX -> (9.6K) -> UCRVM2
 
So, there is an **equal** number of BITNET "hops" from YALEVM
to either UPRENET or UCRVM2.  However, the line-speeds of the
links are different -- perhaps, LISTSERV factors that into its report?
 
The LISTSERV software is only installed at YALEVM, UMDD2, and UCRVM2,
i.e., no other nodes along the path.  Thus, LISTSERV only has a **short**
list of "possible-peers" -- this inhibits the choice of a peer.
 
Advantage of peering:
If UCRVM2 were to become a peer, then each posting to "listname@YALEVM"
would cause **one** file to be sent to 'LISTSERV@UCRVM2', and the **only**
address in the "To:" line would be [log in to unmask]  This would be a
smaller "load" on part of the BITNET network (the links from YALEVM to SERVAX)
than what the current un-peered list places on those same links, i.e., fewer
files will be sent, and each file will be smaller (because the "list-of-
recipients will not be included in each posting).
When the file reaches 'listname@UCRVM2', some copies of the posting are sent
to local subscribers, and then the posting is forwarded to Puerto Rican IDs.