| Many websites now offer a broadband speed test. | | | | events JavaScript is a scripting language that is |
| People are naturally curious to see if they are actually | | | | supported by most modern web browsers. It is used |
| getting what they are paying for. In the UK, many | | | | extensively for applications such as dynamic menus, |
| broadband providers offer connections that fall short | | | | photo albums and pop up windows. Many applications |
| of the expectations that the providers themselves set. | | | | will use simple JavaScript to time the events (i.e. start |
| The almost mythical "up to.." statement is often a way | | | | and end of download) of downloading a file. The |
| for broadband providers to tempt new customers at | | | | problem with this approach is that, as the JavaScript |
| the same time as evading the sticky question of why, | | | | runs on a user's computer / web browser, it can be |
| on average, most broadband connections fall well | | | | subject to fairly high drift based on a user's internal |
| short of the advertised speed. | | | | computer clock. Using simple JavaScript for testing |
| Results from speed tests show that the average | | | | broadband speeds can therefore further contribute |
| speed across all providers in the UK is currently just | | | | towards inaccurate results. Many websites now offer |
| less than 4Meg. Naturally, people are concerned that | | | | Flash based tests which look great. However, users |
| they are getting less than 50 percent of the advertised | | | | need to be aware that, Flash based speed tests are |
| "up to 8Meg" speed and will use on-line speed tests to | | | | usually just a pretty gloss over the top of simple file |
| check their actual connection speeds at various points | | | | downloads, timings using JavaScript. |
| throughout the day to verify the actual speed being | | | | Writing and providing an accurate broadband speed |
| provided. | | | | test requires in depth technical knowledge on |
| Whilst many websites offer broadband speed tests, | | | | networking protocols and cannot be achieved using |
| often the results provided are not accurate. Websites | | | | simplistic JavaScript timing events and file downloads. |
| will use "home grown" tools or applications. Whilst there | | | | Accurate test results can only be achieved from |
| is nothing inherently wrong with "home grown" | | | | professional quality speed test applications. Many of |
| services, in many cases several incorrect assumptions | | | | the more accurate applications for testing broadband |
| are made that mean that the test results are not | | | | connection speeds are highly specialized commercial |
| accurate. | | | | products developed and supported by software |
| The top three common inaccuracies with many "home | | | | companies with expertise in network monitoring. At the |
| grown" broadband speed testers are as follows:- | | | | very least, broadband speed testing applications should |
| Incorrect Assumption 1:- Inaccurate file size calculations. | | | | run on complied language platforms (such as Java). |
| Some applications simply measure the time it takes to | | | | Java allows access to the required elements (such as |
| download a file of know size (for example an image | | | | timers, files and networks) of computer and network |
| file of 500k). The problem comes when doing timing | | | | that are simply not available using simple web browser |
| calculations that require calculating bits per second. The | | | | scripting (e.g. JavaScript). |
| correct size of a 500k file (in bits) is 1024 x 500 = | | | | To get an accurate idea of your broadband |
| 512,000 bits. Many calculations will incorrectly assume | | | | connection speed, it is advisable to:- |
| that 500k = 500,000 bits. In our example, this leads to | | | | - Run tests at several different times throughout the |
| an error of several percent. | | | | day. Three tests (For example:- early morning, mid |
| Incorrect Assumption 2:- Assuming that downloading | | | | afternoon and early evening) should be sufficient for a |
| starts immediately. In timing a download file, many | | | | decent picture of what is happening. |
| speed test applications will start the timing from when | | | | - Use tests from several different test websites. Try |
| a user clicks the "go" button. The incorrect assumption | | | | to find sites with a service located close to you so as |
| here is that the download starts immediately. This is | | | | to avoid further inaccuracies from Internet network |
| not the case as an initial delay is always present when | | | | latency or congestion. |
| requesting any file from a web server. This incorrect | | | | - Use professional tests. Results from many Flash |
| assumption results in additional inaccuracy of the test | | | | based or "home grown" services are fairly basic and |
| result. | | | | only offer an approximation of your connection speed. |
| Incorrect Assumption 3:- Using JavaScript for timing | | | | |