[Tfug] A very nice bandwidth test page...

Stephen Hooper stephen.hooper at gmail.com
Mon Feb 26 00:54:02 MST 2007


On 2/25/07, Jeremy D Rogers <jdrogers at northwestern.edu> wrote:
> Ya, I just got DSL last month from AT&T and once connected, they
> actually pointed me to that speedtest.net as well: 1302 kbps down, 308
> kbps up
> When I'm on campus at Northwestern: 5103kbps down, 4214kbps up
> What is interesting is that I wanted to see how UA compared since UA
> is internet2. So I SSHed in forwarding X and fired up firefox: 209kbps
> down, 4909kbps up. I could imaging this might in part be due to the
> traffic required for forwarding X, but why the disparity between up
> and down. If anything, I would have expected the up to be hurt the
> most since that's the direction of all the data for X forwarding. When
> I wget or apt-get, I usually see around 750kBps or 6mbps down.

I doubt that speedtest.net is an Internet 2 hosted site.  The routing
tables (traceroute) seem to bear me out on that .  So technically your
measurement would be meaningless as you would still be measuring
Internet Apples, versus Internet 2 Oranges.

As to your speed test:  this is a browser based test.  That means that
the speed that code in your browser is running will effect the test.

Does your browser run slower remote than it does locally?  My guess is
that repainting the screen alone will slow down the browser (more time
to repaint means browser doesn't respond to events as quickly (locked
into repainting the screen),  which means that any other events the
browser is doing take longer to respond, etc.).

To prove this to yourself, you could try:
http://celtickane.com/projects/jsspeed.php to see if JavaScript runs
slower over an SSH connection than it does on your local machine...
(hint: it does for me).




More information about the tfug mailing list