Newsgroups: ny.nysernet,comp.mail.uucp
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!uupsi!schoff
From: schoff@uu.psi.com (Martin Schoffstall)
Subject: Re: Ithaca POP for PSI needs PEP : -)
Message-ID: <1991Jun19.174421.7181@uu.psi.com>
Organization: Performance Systems International, Inc.
References: <1991Jun19.015353.717@shaman.com>
Date: Wed, 19 Jun 1991 17:44:21 GMT

In article <1991Jun19.015353.717@shaman.com> jiro@shaman.com (Jiro Nakamura) writes:
>
>In case anyone didn't notice, the Ithaca POP for PSI opened with 2 measly
>v.32 lines....

As "advertised" in numbers-info@psi.com well in advance and implemented
today there are two LowSpeed (300/1200/2400) dialups and two HighSpeed
(V.32, 19.2kbps DTE) dialups in Ithaca.  While this is less than the
25 dialups in NewYorkCity, Ithaca is a smaller city after all....

>I'm using them right now. It is my hopes that more people
>with Telebits will join up in Ithaca, and bully PSI into getting a proper
>PEP line there. :-)

There is little differnce between PEP and V.32 signalling related
to performance on local phone lines.  Given that PEP is inferior
technology for interactive applications and SLIP/PPP, it will
take more than public or private bullying.

>	On another vein, what kinds of throughput are people getting
>with PSI's PEP lines? UUNET only gets me about 900-100 cps (character-bytes
>per second). Since UUPSI goes through a telnet channel (apparently) from
>the POP to PSI central, I don't imagine they get wonderful response
>times (sigh, c'est la vie). 
>

You are not using a telnet channel you are using a TCP Stream on
the official UUCP port (540).

Throughput is related to:

- signalling
- "network" latency +/- spoofing issues
- bandwidth of "smallest" hop within a subnet
- io of source host
- io of destination host

All (save for signalling) are effectively non-deterministic.

Given that they are non-deterministic (and a few other factors)
PSI's services are flat-fee.

Marty
