[log in to unmask] (Edmond McLaughlin) writes:
>Tim Daniels ([log in to unmask]) wrote:
>: I am in the process of working on a mosaic site which will be used to
>: help students design virtual museums. In talking with the campus
>: internet guru he explained to me that my machine, which is off campus will
>: be slow because one of the cable's which connects the machine to the main
>: unix service is not fiber optic cable. It seems that if at any point
>: along the line the connection is not fiber optic the whole transfer
>: process is slowed down. I just thought I would put that out for what it's
>: worth.
>: Tim Daniels
>: TD6560@[log in to unmask]
>For the interested: the speed of the world wide web, or any network
>really, is dependent on two factors: data rate, and latency.
>Data-rate is the speed data can move through a particular medium. Media
>include copper wire (as in your phone line), coaxial cable (as in an
>ethernet network or your cable TV hookup), or fiber optic cable, to name
>a few. The wire connecting your phone to the world is usually copper and
>of a grade to support your standard modem rates: 2400,9600,14400,24400
>bits/sec. A bit is one piece of information: 1 or 0, on or off, +5volts
>or 0volts. The wire in many networks is can be similar to phone wire (but
>of a higher grade) or can be coxial. These will support up to 10mbits/sec
>(10million bits/sec) data rates theoretically, but in practice get
>between 1-3mbits/sec. Soon, Unsheilded Twisted Pair, or UTP, the stuff
>that is like phone cable only much better, will support 100mbits/sec.
>This will be very good. Fiber Optic is capable of even greater data rates
>than that and can move many signals at once across the same cable.
>Latency is the time between a message being sent and a message being
>recieved across a network. This is affected not only by the medium in
>use, but also by the way a message is "packaged", or the protocol used to
>pass messages.
>Mosaic access to the World Wide Web is an information rich feature of the
>internet and requires aconnection providing high sustained data rates as
>well as computer with above average graphics capabilities. (But not that
>high above average). A single graphic image at a medium
>resolution(say 160x180dpi) and 256 colors to appear on your screen would
>require 28000bytes of data to be transferred across the net to your
>computer. If your at a data rate of 14,400bits/sec, you'll need about 16
>seconds for that data to show up on your machine (without any latency)
>and another second or so for the computer to turn it into an image on your
>screen. If your perusing an online image bank or museum, this could get
>tiring.
>The above example is for 14400 baud modem. The access would imporve
>dramatically when going directly through a medium with higher data rates
>( like fiber optic, or UTP). Or using a modem with 24400baud capability.
>Final conclusion: if it's graphical, it's a big CPU and bandwidth
>consumer. A lot of information needs to flow fast and then be
>manipulated even faster to experience any kind of fluid operations when
>dealing with graphic interfaces.
>--
>____________________________
>Ed McLaughlin |
>[log in to unmask] |
>----------------------------
Ed:
I agree with what you have said but I wanted to clear up
a bit what Tim Daniels was saying since I am the guy who told
him about how he is currently connected.
At our site we are in the proccess of replacing a old broadband
system that was constructed in the mid 80's with a fiber backbone.
The building Tim is located is about a mile from the main campus.
It has a 10mb fiber link back to the main campus. However, where
it connects to the main campus is connected to the old broadband
system. The old broadband system is connected in our computer
center to our new main campus fiber ring to be. It kind looks
like this:
Tim ----fiber----Kerr Scott Hall -------Broadband ---+--
|
Off Campus Main Campus +-Computer Center-+
| |
| |
Library -----+ Other
Campus
Buildings
Well it is not a very good diagram.... But hopefully by
Christmas the link that is now broadband between Kerr Scott Hall
and the Computer Center will be replaced by connecting Kerr
Scott Hall to our fiber ring that is under construction. The
broadband is not handling the current load that our campus
ethernet is now tring to push across it due to design and
noise. As soon as the fiber link is in place Tim will have
a nice clean 10mb link to his building and his bandwith will
be much improved. Just today our fiber team was splicing to
add our Computer Science building to the fiber ring. I hope
that they will finish it Friday. To give you an idea of the
performance we get over the broadband when under a load I can
have up to 64% packet loss pinging some of the machines in our
Computer Science building from the Computer Center. Most
buildings have better performance using the broadband but we
are hoping that moving the CS building off the broadband will
allow better performance for the other buildings like Tim's
that will remain on the broadband for a while longer.
Another issue we have here is that our campus is connected to
the intenet via a T-1. Any info our users access that is not
on campus has to cross the T-1. While I enjoy what we have
I wish it was faster.
Just for other info I have been running mosaic on my mac at home
using both slip and ppp via a 14.4 modem. You really see some
delay when downloading inline images but over all it is not bad!
I am hoping to upgrade to a 28.8 modem but as of now our Computer
Center does not support 28.8 modems. And the only reason I am
able to use slip or ppp is that I have a DECstation on my desk
that I started using slip on about 8 months ago. After having
some success I have worked with our telecom group and have been
testing a annex terminal server. It supports slip and ppp and
other stuff like telnet and lat. It seems to work really well.
--
J Martin Moore (704) 262-6274
Appalachian State University
Academic Computing Services
Room 2067 New COB
Boone, NC 28608
[log in to unmask]
[log in to unmask]
|
