What You Need to Know About Internet Speed Tests
Introduction
Internet speed test sites can be useful tools for testing your broadband connection. However, there seems to be a general misunderstanding about how these tools work, and how the results you see relate to the broadband service package you have purchased from Filer Mutual Telephone. This web page will be an effort to explain the technical details of what is going on behind the scenes and hopefully give you a better understanding of how network bandwidth is provisioned and how it is utilized from your computer to a typical Internet web site.
How fast is it, anyway?
All those bits
At it's most basic level, all computer data is represented by binary digits or "bits." A computer only knows two things, "Is the switch on or off?" These two states of "on" or "off" are represented by "1" for on and "0" for off. Each of these is called a binary digit or "bit." If you string enough of these bits together into "words" you can create many, many different combinations that can in turn be combined into larger "sentences" and "paragraphs." When you transmit data from one machine to another across an electronic circuit like a DSL line, the electrical properties of that circuit determine the rate at which each of these bits are transferred.
What's your speed then?
A typical DSL package you might purchase says "1.5M download, 512K upload." What does this mean? The "1.5M" means 1.5 megabits per second. A "megabit" means one million bits. So this represents 1,500,000 binary digits transferred every second. The "512K" means 512 kilobits per second. A "kilobit" means one thousand bits. So this represents 512,000 binary digits transferred every second.
Why different upload and download speeds?
The technical definition of your DSL line is really called "ADSL." This simply stands for "Asymmetric Digital Subscriber Line." Think about what the typical user does on the Internet. You open a web browser like Internet Explorer and click on one of your favorites. When you do this a very small amount of information is sent up to the Internet requesting a web page with the name "www.filertel.com" for example. The result is a much larger amount of information coming from the Internet down to your computer. Since most users download more data than they upload, broadband circuits are built with this "asymmetric" property to conserve resources. For example, why should you pay for 1.5M for uploading if you never use more than 512K?
Let's Test It!
Enough of the boring background information. Let's take a look at a typical Internet speed test site and see how it works.
This is a typical speed test site that you can visit at www.speedtest.net. When you open this web site, it is smart enough to determine that you are in southern Idaho based on your IP address. Notice that it has already selected "Idaho Falls" as the "recommended" server to test with. Why? Well, what happens when you start the test is pretty straight-forward. A fixed amount of data is transmitted to your computer and measured against an accurate clock to determine the rate at which the test data was sent. Then, another bit of test data is transmitted up from your computer back to the testing server and measured. The result is displayed as your download and upload speeds. All well and good right? Not so fast! It turns out that it really matters where this testing server is located in relation to your computer when speeds are measured in this way. The number of "hops" it takes to get to this server and back to your computer has a big effect on all those bits flying back and forth across those wires. Suffice it to say, the closer the server is to you, the more accurate the results will be.
How many hops?
Even though Idaho Falls may look like it's just around the corner on the above diagram, it may not always be the case. Actual pathways that network traffic traverses may appear counter-intuitive at first glance. To illustrate this, you can use a Windows command-line tool called "tracert." When you run this "trace route" program on a destination IP address, it will output a list of all routers between you and the destination. Look at the following example:
The IP address of the Idaho Falls "speedtest.net" server is 69.20.128.69. When you do a traceroute to this address from Filer Mutual, you can see it seems to take quite a long path. Why is this? The following diagram will help to make sense of what you see above:
Click HERE to see a larger view of this diagram.
Note: Filer Mutual can only guarantee your link speed between your computer and hop #1. Anything beyond #1 is treated on a first-in/first-out best effort service. However, the network between #1 and #4 is all on a gigabit ethernet infrastructure so there is plenty of bandwidth to prevent any bottlenecks without the need for any special traffic shaping.
The first 4 "hops" you see are all within the Filer Mutual network. This is indicated by the blue rectangle area. Specifically, hop #4 is the Filer Mutual Router that connects to the Main Internet Feed coming from Syringa Networks. Hop #5 is the Syringa Networks Router on the other side of that link in Boise. The link between your computer and hop #1 is where we configure the bandwidth for your broadband package. The link between #4 and #5 is where we provision bandwidth with our service provider to accommodate the needs of all subscribers.
Inaccuracy in speed test results increases with hop count.
If you look closely at the traceroute output you can see the path from Filer to Idaho Falls goes to Boise then Portland Oregon, then Seattle Washington, then a couple more unidentified hops then finally the Idaho Falls speed test server after 17 hops! Filer Mutual has absolutely no control of how Internet traffic is treated after it leaves our area of responsibility between #4 and #5. This is why even though speedtest.net recommends Idaho Falls as the closest test point it is still too far away from a network routing perspective to provide 100% accurate results all the time.
Now you can see why the above results are so poor. This test was performed from a DSL line configured with a 2M download/512K upload profile. After hop #4, your network traffic basically has to compete on a first-in/first-out basis with all other Internet traffic traversing 12 additional router hops spanning hundreds of miles. This is what is meant by best effort level of service.
Syringa Networks Speed Test is the best choice.
Syringa Networks is our Internet provider. They have a speed test server installed on their network that is free for you to use. You can see it is located just off of hop #5 in the above diagram. In order for you to accurately test the speed of your broadband connection, this really is the best option. You are still on a shared infrastructure that contains network traffic for other Filer Mutual patrons and other Syringa members, but you eliminate the extra overhead of hops 6 through 17 as in the above example. We have a link for the Syringa speed test right on this website. Just click the link in the "Online Tools" section to the left:
Then just click on the "Boise" or "Idaho Falls" link to start the test:
The above test was performed from the same computer as the first one but this time with dramatically better results.
We here at Filer Mutual sincerely value your patronage. We hope this technical discussion was helpful in aiding your understanding of what goes on "behind the scenes" and how your Internet bandwidth is provisioned and managed. We always strive to provide you with the best possible Internet experience and we are always seeking new ways to increase the quality of the services we provide.
You can always call our technical support line at 326-3700 if you have any questions about your Internet service.