One of the best things about signing up for the latest and greatest broadband service is the speed, but how can one really measure the speed of their broadband connection? After all, is there really a difference between a 15 Mbps per second DSL connection and the performance offered by a 15 Mbps cable modem service? Certainly not, if one listens to people complaining on various forums around the Internet. The answer to that question might depend on a great number of factors, and surprisingly, there are two different ways to measure broadband performance. Before looking at the two different kinds of tests, it is worth looking at a few key terms that many people overlook.
A Few Helpful Definitions
Downstream : Downstream is the speed at which a computer can download data through its internet connection. Examples of when downstream performance comes in handy include downloading large files and watching streaming video.
Upstream : Upstream is the measurement of speed leaving he computer, bound for some destination on the Internet. Examples of situations where high upstream speeds come in handy include updating blogs, posting pictures to sites such as Flickr, and adding videos to one’s YouTube account.
Ping : Ping is a measurement of how long it takes a piece of data to travel to a specific destination and return. Low ping times are important for video conferencing, voice communications such as VoIP (Voice over Internet Protocol), and gaming. In fact, low ping times are essentially for gaming, but no less important for real-time communications.
Synthetic Tests and the Shape of Things to Come
One of the most common ways that people measure broadband performance is to visit sites that offer synthetic performance tests. These tests are called synthetic because they measure abstract tasks that are often simulations of real-world tests. Almost everyone has heard that there is a significant difference between theory and reality, and this same gap may appear when trying to use a synthetic benchmark for a broadband system. That is to say that real world usage patterns may produce different results than a synthetic test, most often on the lower side.
The reason for this could be attributable to Internet traffic shaping, which is essentially a way of saying data-prioritization. Some data is very time sensitive, such as real-time voice and video conferencing data, but less so in other applications such as uploading and/or downloading large files. For example, if a few data packets containing video and/or audio information are delayed in a real-time conference, the results would be notable distortion. On the other hand, people may not notice a few hundred bytes per second fluctuation in a large upload or download; in fact, such fluctuations are quite normal.
The reason that such fluctuations are normal is the aforementioned traffic shaping. The further the data is from where it is going, the more networks it has to go through, and thus the more chances are that the data will be delayed due to traffic shaping. This makes trusting ping times particularly difficult when it comes to synthetic tests, because they are essentially a controlled environment. Upstream and downstream speeds may be more reliable, but they too change over the course a day. Why? Because people use the Internet more during some hours than others, and many upload/download tests used to measure upstream and downstream are impacted by the additional traffic.
Real World Testing
The best kind of testing is probably that of real world testing. After all, who buys a connection only for the purposes of running synthetic tests? If things that used to be sluggish seem fast, then chances are good that the new broadband connection is better suited to one’s needs. Real world tests are rarely easily and unequivocally measured, but they are the tests that matter. Unfortunately, the real world often makes for unreliable tests; one day the connection will feel blazing fast, while the next day it might not seem so quick. There are simply too many factors to consider in many real world tests, but that does not make them any less important.
Additionally, some benefits of faster connections are only apparently when and if multiple applications are running and using the Internet at the same time. For example, ultra fast connections might be able to handle downloading a full season of Lost at a reasonable ratewhile not causing tangible increases in ping times. Slower connections might make downloading large videos while playing a game a high-ping experience that results in a bad experience. Worse, the download speeds are likely to be negatively affected as well.
On top of all this, measuring broadband performance in a real world setting is often subject to more influence by the number of users on the Internet that are using servers between the test site and where data is going to/coming from. In a synthetic test, all of the data is coming from and going to a single destination in most cases, but that is rarely the case in the real world. People are downloading video files and software patches while uploading photos to Flickr, or streaming video to a media box while playing World of Warcraft. The bottom line is that real world speed tests offer so many factors, but that does not mean that they cannot be tested. It just means that the results are based on one’s own usage patterns, schedule, and the schedule and usage patterns of other groups of users that create a sort of ‘Internet interference’ that results in slower data packet processing.
Summary
Measuring broadband performance can be done in many ways, but real world tests are often more important than synthetic tests for any number of reasons. That does not mean that synthetic benchmarks are unimportant, but they are ultimately less important than they might make themselves seem. The best way to measure broadband speed is simply to give it a try. Unfortunately, that can be quite expensive, and the apprehension borne from this quandary has given root to all manner of myths regarding why DSL is better than cable, or why cable is better than DSL, or even why people should just go back to dial-up.
