Taking the Long Way Home

Today I was benchmarking my internet speed using Speakeasy's speed test.  One of the reasons for this is that I've noticed some inconsistencies with Telus' network speeds.  Anything that's outside of a 250 km radius to me tends to cap out at about 1100 kbps, while sites as close as Seattle can send at almost my full 3.0 Mbps.  To prove the point, here is the best speed I managed to get from Seattle, Washington today:

Seattle Run

My test scores from Texas, New York and Illinois averaged at just below 800 kbps.  I tried these tests during different times of the day, and on different days with no real difference in speed.

Digging a bit deeper, I decided to do a trace route for these sites to see just how the data is routing through the net before reaching my computer.  From Seattle, data hops through an average of 12 servers.  From Austin, TX, data hops through just over 30 servers.  From Buffalo, NY, data hops through Montreal, then what appears to be every city with a population higher than 100 people straight through until Vancouver.

I wonder if some network sultan decided to take some revenge on Telus, because I can't believe that something like this would be missed.

For the last few months I've been struggling with speeds moving between here and Japan.  Reiko and I often talk on MSN with our webcams, and we're regularily struggling with constant disconnections and network pauses.  I've pulled my hair out thinking that it might be my hardware, or Reiko's hardware, or something between our different computer setups … but all of this checked out.  But after doing these tests (not only for speed, but for routing), I have something else to consider.

Broadband isn't just about speed, but reliability.  People pay the extra for broadband internet because it's supposed to be both faster and more reliable.  Telus has already failed miserably with reliability, considering how their DNS servers are about as useful as a 100 yen coin in Istanbul.  Perhaps this routing is just the tip to something else I need to examine.

I know that data will hop from server to server as it travels from one part of the globe to another.  I'm just surprised that it has to jump as often as it does.  With Shaw, I would often see no more than ten hops if I was getting something from anywhere in North America.  Of course, depending on who Telus has angered, they may be powerless to prevent this.

A friend of mine at work has often threatened to have someone route all our external traffic through Uzbekistan (in jest, of course).  Perhaps this is what happened to my long-distance packets, too.