Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uwm.edu!linac!pacific.mps.ohio-state.edu!zaphod.mps.ohio-state.edu!casbah.acns.nwu.edu!news
From: news@casbah.acns.nwu.edu (Mr. News)
Newsgroups: comp.dcom.telecom
Subject: Re: Prodigy: What Does Sears Do? What About IBM?
Message-ID: <telecom11.131.4@eecs.nwu.edu>
Date: 18 Feb 91 11:03:01 GMT
Organization: Foretune Co., Ltd.  Tokyo Japan
Lines: 52
Approved: Telecom@eecs.nwu.edu
X-Submissions-To: telecom@eecs.nwu.edu
X-Administrivia-To: telecom-request@eecs.nwu.edu
X-Telecom-Digest: Volume 11, Issue 131, Message 4 of 5
Originator: telecom@delta.eecs.nwu.edu


hayes!tnixon@uunet.uu.net (Toby Nixon) writes:

>By the way, I might comment on why it is that Prodigy was so concerned
>about high levels of email traffic.  Email on Prodigy CANNOT be staged
>on the nodes -- it HAS to go all the way to White Plains for every
>screen.  Heavy email traffic saturates the network, which can bring it
>to its knees very quickly.  Email was intended to be an occassional
>convenience feature, not a primary use of the service.

Well, I won't bother to comment on the utter stupidity of the
conclusion that email was going to be a minor feature of a nationwide
consumer computer network, even absent of the fact that email was
encouraged as a way to deal with the "editing" of messages posted to
the Prodigy Forums.  Obviously, Sear's massive market research didn't
include logging on to any of the other online services ... ;)

However, I do take issue with the statement that "Email on Prodigy
CANNOT be staged on the nodes."  Thats simply not true.  It could very
easily be staged.

Since Prodigy knows which local host a user last logged into, and
since it is probably a 95% probability (or more) that he will log into
this same host on his next session, that host can be declared to
contain his "mailbox", and mail can be catched there in anticipation
of his next access.

Insofar as "instant" delivery of mail is concerned, the only time
Prodigy has to be concerned with priority delivery of mail from the
central host to the local cache is when the user is actually on the
system and in the email application; otherwise, mail that is sent to
the user can be sent to the cache during times of low system
utilization.

Of course, the five percent of the time when the user changes local
nodes doubles the overhead for pending mail, as it has to be sent from
the old local node through the host to the new one (I assume that
Prodigy, being cheap, doesn't want to spring for the extra disk space
to duplicate email on the host and the local nodes).  However, since
this only happens one time out of twenty, and the other nineteen times
the cost at the time of reading is near zero, you are looking at about
a 80% decrease in the "cost" of sending email, at least.

Finally, PRODIGY can easily deal with the huge mailing lists by not
transmitting an identical copy to each recipient, but rather just one
copy to each node that has recipients for it.

Why they did it the way they did is a story I for one would like to
hear.


Robert J. Woodhead, Biar Games / AnimEigo, Incs.   trebor@foretune.co.jp