A big thanks to the 11 people who responded to my call last week to do a speed test on their Internet connection. This is not a bad response as some reported that my e-mail address was missing the 'at' symbol.

The results are mixed. Some achieved close to what they paid for, while others were below the mark as you can see below. As background, here is a small recap for those who are not in the loop. Last week I wrote about the quarterly State of the Internet report issued by Akamai. The company reported that average speeds to Akamai's servers from Bermuda was 1.5 Mbps in the third quarter of 2009, down 4.9 percent from Q2 2009, but up 2.5 percent from Q3 2008.

I noted there is always a difference between the maximum speed offered by a service you pay for and the actual speed achieved on a given day. This difference is due to a number of factors, including the load on the network. The number provided for any given service is the maximum speed you can achieve if all was perfect.

Out of curiosity I asked readers to send me in their readings using a speed test site (www.speedtest.net) and a ping test (www.pingtest.net). Paul Coleman, the vice-president of marketing at Logic Communications, also wrote in with a helpful e-mail to further elaborate on what a speed test means. It turns out that testing Internet speed is harder than it sounds, he says.

I quote directly from his e-mail. "The results of the tests depend on the following:

1. Amount of network activity going on, on a computer at the time of the test. Modern OSes [operating systems] can do a lot of background tasks, such as virus signature updates, critical patch updates, Microsoft Customer Feedback automated uploads - and the list goes on. If any of these are running when the test is performed the results will be affected.

2. The location of the speed test site. Some sites are located in Bermuda (the one you recommend in the article is hosted by TransAct and is physically located in Bermuda; there are options on that site to pick different servers in different locations). Testing to a site in Bermuda will not test the ability to access overseas Internet destinations.

3. 'Upstream' congestion. Upstream congestion refers to network congestion between the edge of the ISPs network and the ultimate destination on the internet. This is related to #2: if a speed test site is in Chicago and the ISP network ends in Miami or New York, upstream providers will be used to move the data from NYC to Chicago. The amount of congestion on those links will affect the results.

4. The configuration of the speed test sites. Different transport layers (the access providers in Bermuda: CableVision, BTC, and M3 easyConnect) use different technologies with different characteristics. Depending on how the speed test site has been optimised, equivalent networks would get different readings on speed test sites configured differently.

5. ISP/Access Network maintenance. While usually short in duration, certain types of network maintenance activities that the ISPs and access providers do can affect speed results."

As you can see, this clarifies a lot. But it also shows that Internet Service Providers (ISPs) should put disclaimers when quoting service levels and speeds, as their posted speeds are confusing (and perhaps misleading) to customers.

One should not expect customers to have to figure this technical stuff out. As with other consumer goods and services, they want what they paid for, and when they do not get it, they want to know why. Instead of having this kind of frustration build up, disclaimers should be posted.

Lastly, I did not know the location of the speed test site mattered. The fact that the one I pointed to (as recommended in a Computer Society of Bermuda newsletter) is in Bermuda works for me, since it establishes close to a 'perfect' testing situation for the speed test site.

Steve Moffat, operations director at Optimum IT Solutions, also wanted to point out that the recommendation to use www.speedtest.net and www.pingtest.net is "not a true reflection on Bermuda ISP Internet speeds".

He adds: "ISP Internet speeds are only guaranteed between the end user & the ISP's equipment, therefore using either of the above websites give spurious returns as Bermuda ISP's cannot and do not guarantee the Internet speed once it leaves the ISP's equipment. I would be grateful if you could point this out in your next article and update the CS's information with the real truth."

Moffat put a smile at the end, but really it illustrates my point that ISP's should put a disclaimer so that their customers know the "real truth".

Now for the results. Tester #1 has a 3 Mbps connection with Logic and was getting a download speed of 2.76 Mbps (megabytes per second), an upload of 0.58 Mbps and a ping of 88 ms (milliseconds) on testing. Tester #2 has a 4 Mbps connection (Logic), and achieved a download speed of 2.81 Mbps, upload 0.59 Mbps and ping of 56 ms. Tester #3 did several tests at different times and has two accounts. I assume this data is from his 4 Mbps account with BTC given the download speed results: he got test results in the range of 3.15-3.30 Mbps for download and 0.57 Mbps on upload and a ping of 45 ms.

Tester #4 has a 3 Mbps service (Logic), download 2.94 Mbps, upload 0.57 Mbps, ping of 88ms. Tester #5 has a 1 Mb service (North Rock), achieved 1.07 Mbps download and 0.58 Mbps for upload. His ping test was 34 ms with no packet loss (an important parameter that I neglected to ask for).

Tester #6 has a 3 Mbps connection (Logic), achieved a download of 1.39 Mbps (did not report upload). The ping test was 59 ms with no loss. He added: "I run a speed test on my system probably once a month - I don't think I have ever been above 2.0 Mbps, yet I am paying for 3 Mbps - and have never had a satisfactory answer from either Telco or Logic as to why that would be the case!"

Tester #7: 3 Mbps account (Logic), download 2.72 Mbps, upload 2.51 Mbps, ping 79 ms. Not bad on the upload! Tester #8 has a 3.5 Mbps connection with Transact and achieved a download speed of 3.2 Mbps, an upload of 0.57 Mbps and a ping of 72 ms with no loss.

Tester #9 has a connection of 4 Mbps (BTC), a download of 0.39 meg, up-load of 0.11 and ping of 104 ms, adding up to a very poor connection speed. He reports he is writing a formal complaint to BTC "as my verbal complaints have been neglected. I live on the south side of Harrington Sound, which, according to the BTC technicians, has not been upgraded by BTC for many years".

I assume BTC will respond? Let me know.

Since I have very little space left, I can report that the remaining other two testers had speeds reasonably close to the service they pay for, as did many who did the test. Others, such as Tester #9, might either have problems at home with their hardware, or might have a legitimate beef.

I am sure I will get some answers from the ISPs in Bermuda as to these results.

Send any comments to elamin.ahmed@gmail.com