Have you ever wondered about how fast your internet connection REALLY is? If you are using a dialup modem, do you just take as gospel the connect speed that your operating system reports? Or maybe you use one of the online speed tests such as the one at ZDNet to see how fast your connection is? If so, have you ever bothered to stop and do the math to see how accurate those numbers are?

Just for grins, let's work through some of those numbers. Let's say you've got a 56K modem and Windows is reporting a connect speed of 52000 bps. Since that measurement is in "bits per second", let's convert that to something more meaningful -- like "characters per second", since "characters" is something that we can understand. To do that we just divide the "bits" by eight, since there are eight bits to a character. So your reported speed of 52000 bps is really about 6500 characters per second (cps). Or so Windows would have you believe.

Now let's assume that the average web page consists of about 42,000 characters (actually that's not all that bad of an assumption). So let's do the math here -- supposedly your connection is running at 6,500 cps, so the average time to download a page should be 42,000 divided by 6,500 or about 6.5 seconds. Clear out all your temp files and then click around to some of your favorite sites and time how long it takes for them to fully display. Bet it's a lot longer than 6.5 seconds.... in fact, if you are using a 56K modem, you are probably averaging about 14-15 seconds per web page (see Keynote.com for more stats).

And where the math really gets interesting is for high-speed "broadband" connections. Say you have a 384K DSL connection. Do the math on that one.... that's 48,000 cps, which means that the average web site should display in less than one second. If you've ever been on a DSL connection you know how far from reality that is!

So what's going on here? The "problem" is that just because your connection is capable of receiving data at a rate of say 52Kbps, that doesn't mean that you are going to have a steady stream of information coming to you at that rate. What is happening is that you are running in to traffic on the internet that is causing your particular web page request to get delayed at various points in between your computer and the web site that you are accessing. Thus your throughput -- how fast data actually gets to your computer -- is going to depend on a lot of factors over which you have no control.

Now you may have seen some web sites that purport to measure your true "throughput" speed, such as the MSN Computing Central site mentioned above. But the problem with those type measurements is that about all they tell you is how fast you can download one particular file from one particular location. If you spend the majority of your time on the internet just downloading files, then those types of measurements may be somewhat accurate for you. But if you spend the majority of your online time just "surfing" the net, then those measurements are pretty much useless. The reason for that is because web pages typically consist of a lot of small files that must be downloaded to your computer, not one big file. And quite often, those files reside on more than one computer, which has the potential for slowing things down even more. And as we demonstrated above, if you think these types of measurements are accurate, then run a test at one of those sites, then do the math on how long it should take you to download web sites and compare that to what you really see happening.

INETBENCH is designed to give you an accurate way of assessing what your connection throughput really is. It does this by actually downloading web pages from the internet using the exact same windows interface that your web browser does. Whatever delays and overhead that you encounter using your browser will be accurately reflected in the statistics that INETBENCH produces. You can run benchmarks at different times of the day to see exactly how much faster or slower things are depending on the time of day. Or if you have multiple dial-up options, you can test them to see which performs best for you. Or if you happen to have a high-speed connection like DSL you can see just how much faster it really is compared to your dialup.