[8704] in Commercialization & Privatization of the Internet
One-way Internet?..no, it's "Asymetric"
daemon@ATHENA.MIT.EDU (Mark Costlow)
Wed Dec 1 00:51:28 1993
Date: Tue, 30 Nov 93 19:35:59 MST
From: bbx!kitsune!cheeks@cs.unm.edu (Mark Costlow)
To: bbx!mimbres!ssf-sys.dhl.com!Paul.Rarey
Cc: com-priv@cs.unm.edu
In-Reply-To: <9311301643.ZM2276@maverick.ssf-sys.DHL.COM> (bbx!mimbres!ssf-sys.dhl.com!Paul.Rarey)
On Tue, 30 Nov 93 16:43:35 -0800,
Paul Rarey <bbx!mimbres!ssf-sys.dhl.com!Paul.Rarey> said:
Paul> On Nov 30, 16:26, Richard Smith wrote:
Paul> }
Paul> } Do any of the current network providers include bandwidth guarantees?
Paul> } How do they do that? I can imagine scenarios where even a dial-up
Paul> } connection would slow down due to traffic/use on the host. What does
Paul> } it say in *your* fine print?
Paul> How does one measure CIR on FrameRelay offerings today? Or is that
Paul> just a farce also?
I hope OARNET doesn't mind me using them as an example.
I've wondered about statements such as the following (from OARNET's product
literature):
Low Volume T1: A Low Volume 1.5Mbps (LV-DS1) connection
is in technical terms the same as a T1 (1.5Mbps) attachment. The
peak utilization of the physical connection is what determines
when the connection is a Low-Volume 1.5Mbps or a regular T1
(DS-1) connection.
If the peak utilization measured on a daily basis exceeds (a) 20%
of the capacity of a 1.5Mbps (DS-1) circuit for more than three
days, which need not be consecutive, in any three consecutive
months, or (b) 20% of the capacity of a 1.5Mbps (DS-1) circuit for
more than three days, which need not be consecutive, in any six
months in any twelve months contract period then the connection is
considered to be a T1 Service.
Peak utilization means the average of the number of bytes delivered
to, and accepted from, a site ((delivered + accepted)/2) during the
15 minute period with the highest traffic between 9:00AM and 6:00PM
Monday through Friday, as a percentage of the total capacity of a
full duplex T1 (DS-1) connection.
The thing that I wonder about is how one throttles throughput. If you've
physically got a T1 line, but you're allowed to consume more bandwidth (at
your discretion), what keeps a user from consuming maximum bandwidth all
the time? If you FTP to some place reasonably nearby (bandwidth-wise), and
"get" a big file, what keeps the unmodified ftp and ftpd programs from
consuming as much bandwidth as is available? As far as I'm aware of, there
aren't any mechanisms to deal with this in most OS vendors' network
software. Are there CSU/DSU's or routers which can perform this function?
Mark
--
cheeks@kitsune.abq.nm.us or ...mimbres.cs.unm.edu!bbx!kitsune!cheeks
