[Ham-Mac] Internet Speed Test (How good is good?)

[Mike Carter]

Just a couple points about ADSL and these online "speed tests" in case 
anyone's interested...

Tests like these shouldn't be used to gauge how good your connection 
is. The best thing they're useful for is checking your particular route 
through the Internet to their test site. Remember, these are 
point-to-point tests across the internet. Everyone will have different 
results since everyone has a different path through different routers 
to get to the site. This makes comparison of the results baseless.

For ADSL users, the true indicator of how good your connection is a 
combination of your distance to the CO (phone company Cenral Office),  
the type of switching equipment your telco uses, quality of the copper 
underground and the quality of your ISP's interconnects to the 
backbone. The theoretical maximum throughput that ADSL can carry over 
it's pair of copper is 8Mb downstream and 1Mb upstream. Most 
residential users will never see this, as the throughput is throttled 
at the DSLAM into different tariffed allotments. Your distance from the 
CO dictates the maximum speed of your DSL, which has a maximum copper 
distance of 18,000 feet. So basically, the closer you are to the CO, 
the better your rate *can* be (ask the telco when they install what the 
line distance to the switch is).

The most popular basic tariff is 384/128 (downstream/upstream). At 
384Kb, you can expect a theoretical max download rate of 48KB/s 
transfer times. Many things stand in the way of ever seeing that max 
rate: performance on the Internet, the connection speed of the remote 
host, latency across hops, packet fragmentation, interleaving in the 
DSLAM, etc. I use a factor of 85% of maximum as a base when analyzing 
performance.

So, when you go to these "speed test" sites, you are only getting very 
specific throughput information based on your particular route to their 
server, at that particular time of the day. The best measure of the 
health of your connection is to know what to expect for your particular 
installation, and to make your assessment using several different sites 
at different geological locations in order to arrive at an average. And 
finally, to know that conditions are always changing, and you'll always 
see better and worse results.

Hope this sheds some light. :)

73,

Mike - K6MEC
Senior Network Engineer/CCNA

On Wednesday, Sep 17, 2003, at 21:02 America/Los_Angeles, Chuck 
Counselman wrote:

> harris ruben wrote:
>> For a real surprise, try THIS test side:
>>
>> http://speedcheck2.optonline.net/speedcheck/speedcheck.html
>> ...
>> There is a 3X difference in speed between the 2 sites!
>
>
> Not here, not now.  I tried it twice and got about 1200 and about 1100 
> Kbps; whereas the computers4sure test yielded slightly below 1100.
>
> -Chuck, W1HIS
> _______________________________________________
> Ham-Mac mailing list
> Ham-Mac@mailman.qth.net
> http://mailman.qth.net/mailman/listinfo/ham-mac
>