Three guys from Time Warner showed up at my door. Nobody was toting a straight jacket, so I figured they weren't going to try to haul me away.
Al Fasoldt's reviews and commentaries, continuously available online since 1983


Road Runner Wideband speed tests: Wireless PCs slow down a combination wired/wireless network

Expert: Actual Internet speed is 'almost guaranteed' to be less than advertised

July 25, 2010

By Al Fasoldt

Copyright © 2010, Al Fasoldt
Copyright © 2010, The Post-Standard

   The folks at Time Warner were understandably concerned when I reported my findings on the speed of the company's new Wideband Internet service. (The article ran in early June. You can find it at www.technofileonline.com/texts/tec060610.html.)
   My tests showed much lower speed than Time Warner advertises for Wideband. Rather than the rated speed of 50 megabits per second (mbs), I was getting results in the 30s, with some tests showing even slower speeds.
   After that column ran, three guys from Time Warner showed up at my door. Nobody was toting a straight jacket, so I figured they weren't going to try to haul me away. I showed them my computer setup.
   With six other eyes following everything on my screen, I repeated the tests I had run previously. The results were the same. They had already checked the cable coming into my house; it was good. And the cable coming down my road. It was good, too.
   But what was wrong? Is it possible that Wideband isn't actually 50 mbs? Or was there something wrong in the way I did my tests?
   The Time Warner trio promised to get back to me and left with numbers scribbled on notepads.
   When they called me a few weeks later, they unraveled the mystery. Here's what was amiss in my tests:
      -- I had wireless computers connected to my home network when I ran the tests. There was no wireless connection to the computer running the tests, but three other computers were connected to the same network by wireless connections. They dragged the router's speed down to match the slowest wireless connection.
   In other words, the router, which has wired ports along with wireless capability, was not able to reach 50 mbs on the Ethernet connection to my test PC as long as even one other computer was connected wirelessly. (Wireless speeds are much slower than most wired network speeds.)
   I suspect that most users, even those with a great deal of experience, are unaware of this problem, which affects all routers.
      -- I was running my tests on servers outside Time Warner's own lines. Time Warner's claim of 50 mbs applies only to speeds between one TW server and another -- between the server in Syracuse, which I'm connected to, and the one in, say, Rochester or Albany.
   Before you complain that this is an unfair way to advertise Internet speed, I should point out that it's the only way to advertise Internet speed. Time Warner has no control over the quality of other servers and their connections, nor can it do much to alleviate congestion on the Internet.
   Time Warner probably should tell prospective customers that real-world speed won't match the claimed 50 mbs, but this shows an industry-wide reticence that's not limited to TW.
   When I shut down the wireless connections on my network and tested again using Time Warner's servers, I got 50 mbs, give or take a few megabits. One time I even got 53 mbs. My "off-site" tests showed improved speed, too, but I never got close to 50 mbs.
   But that's OK. At least now I know how to boost the speed of my wired PCs and Macs, simply by shutting down my wireless computers. And 50 mbs? For those rare times when I'm viewing Web pages on a Time Warner server, I know I can fly.