… broadband speed tests are not all created equal. And Broadband speed is not the simple concept we would like it to be.
The speed of our broadband connection matters.

It matters to retail service providers (RSPs) in terms of promoting their services and in up-selling customers to a more expensive service.

It matters to us, as customers of RSPs, in terms of service quality and in terms of getting what we pay for.

Service quality is about being able to download web content quickly.  For example, we expect to be able to stream Youtube video without having it stutter and stop.  And we expect to use Skype without having it pixelate to a muddy morass of unrecognisable greyness.

speedtestIn terms of getting what we pay for, our reasonable expectation is that, if we pay for a 30Mbps down and 10Mbps up connection, we will get close to those speeds.  Some test results show that that is not always the case.

Truenet provide a service to the Commerce Commission that monitors and publishes broadband performance data for homes and small businesses.

Truenet’s December 2013 report says “ADSL file download speeds continue to achieve better than 90% of advertised speed”.  That is an outcome significantly different from Truenet’s measurements on my Vodafone Rural Broadband Initiative (RBI) connection.

For that RBI service, Vodafone’s download speeds are consistently around 50% of the promised 5Mbps.

So what is going on?  Am I getting the service I have paid for?

There was a time when Truenet reported download speeds in excess of 4Mbps and I was happy to be getting what I paid for.

When Truenet’s reported speeds declined to under 2Mbps for my connection, I decide to investigate.  Since starting to ask questions, the Truenet-reported speed increased to around 2.5Mbps.

The starting point for my investigation was to measure the connection speed using another broadband speed test service.  The website speedtest.net is used by most if not all RSPs because they use a globally recognised methodology which allows RSPs to compare statistics amongst their competitors.  This service produced speed results of 5 Mbps consistently at any time of day and to any national server with international servers returning similar results.

Such a consistent speedtest result was suspicious, so other speed test sites were checked.  They confirmed speeds similar to those of Truenet’s and around half of those of speedtest.net.

This raised the question of whether Vodafone was filtering data traffic to give priority to speedtest.net services.

That question was posed on geekzone.co.nz.

The responses were an emphatic NO.  Which I accept but that does not answer the questions of what has changed to halve speeds over 2 years and why is there such a big difference in speeds as measured by different speed test services.

In response to my queries to Vodafone and to Communication and Information Technology Minister Amy Adams, they each believe that the RBI service obligations are met on the basis of my 5Mbps speedtest.net results.  It seems that Vodafone also measure broadband speeds via probes installed throughout their network and with their customers.

It transpires that broadband speed tests are not all created equal. And Broadband speed is not the simple concept we would like it to be.

There are a number of factors that influence speed test results and with the Telecommunication Carriers Forum developing a Broadband Testing Methodology, this is a good time for a user perspective to be considered.

Next week I shall explore answers to the two questions posed.  For background information on broadband speed tests, and to complete a survey of your broadband speed using a range of test services, please visit ruralconnect.org.nz.

Actually, what is broadband speed?

To explain broadband speed, consider my need to get 100 litres of water to a stock trough 1,000 metres away from the main water tank over an existing 12mm pipe.

Think of the main water tank as the web server and the 100 litres of water as the file to be downloaded to your computer (the stock trough) over the broadband connection (the 12 mm water pipe).

When I turn the tap on at the stock trough, there is a small delay before water starts to flow. This delay is like the latency of the broadband connection.  Latency (in milli seconds = thousandths of a second), sometimes referred to as the ‘ping time’, is a major determinant of actual download speeds and is something that is not often mentioned by RSPs.

For a more detailed exploration of what latency actually means to broadband users, check out …coming soon…

The size of the water pipe equates to the bandwidth of your broadband connection and is often the value quoted by RSPs as the broadband speed. It is more accurately referred to as the ‘headline’ speed. This speed is always higher than the actual speeds you will get because it excludes latency (the delay before your broadband signal gets to its destination) and other overhead data needed to make the connection work.

When the water starts to flow, the flow rate (litres/second which is equivalent to broadband speed in Mbps) is slower than desired. It is simply taking too long to fill the water trough.

So I can do a number of things.

First, I could change the pump to one with a higher capacity. This is like changing the local network or the computer that serves the web page to one with a higher speed. It will increase the flow rate only if the limit to how fast water can flow through a 12mm pipe, has not been reached. The latency of the pipe is fixed and also slows down the actual speed of the pipe.

Or I can add to the pipe by putting a second 12mm pipe alongside the first one and connecting them in parallel. This is like moving from an ADSL connection to a VDSL one – the RSP simply binds two copper wires (pipes) together to double the capacity of the pipe to your home.

That improvement satisfies me for a while but when I increase the number of stock needing water (equivalent to the kids returning home or signing up to a new video download service), that double pipe is no longer sufficient.

So then I can replace the 12mm (copper) pipe with a fatter 25mm (fibre) one. Doing that increases the bandwidth of the connection and also reduces its friction and so reduces the delay (or latency). The two together, result in an increase in the water flow rate (or speed) of the connection.