Speed Tests are only half the story

We've probably all done it: jumped onto www.speedtest.net to check our connectivity but what is it actually telling us?

Broadband speed tests do a number of things when to activate them.  Most send a "ping" message to your selected server.  This measures the latency or delay in milliseconds.  They are also using your IP address to determine your location and most allocate the most local server to you. Then to program sends and receives data to measure the download and then upload.

Broadband Speed or Local Wi-Fi speed?

Here's the first key issue.  If you are running a test from a phone or Laptop and you are using Wi-Fi you are measuring the speed of both your broadband and your Wi-Fi.  Your Wi-Fi might be slower than your broadband especially where you have unseen interference from, for example, electrical equipment especially microwaves.  Microwaves work at 2,450 MHz which is VERY close to one of the frequencies of some Wi-Fi. 

So, to be sure you are measuring your broadband speed you first of all need to be connected to the Router and ideally by a fast Ethernet cabled connection.  This would mean the slowest part of the chain is your broadband.

In short you are only measuring the slowest bottleneck or link in the chain and that might well be due to your network.

Which speed test Server?

When the speed test software starts it takes your geographic location and chooses a local server to you.  This is to give you the highest possible result to show you your maximum broadband speed.

However, there's a problem with this.  As a Virgin customer (patiently waiting for Truespeed) speedtest.net selects Virgin Bristol by default.  That server is about 6 miles away from me.  Great, so my test shows me good speeds but the data never leaves the Virgin network. Therefore, I am only measuring the maximum speed between Virgin and me.

In the real world internet

In the real world, it's highly unlikely I want to connect to anything within the Virgin network and my real world speed might be bottlenecked elsewhere.  It's been widely publicised that many broadband providers especially the bigs ones have been struggling since the Covid19 pandemic with their core networks running at capacity.

So should Virgin's core network be maxed out at busy times with say YouTube, iPlayer or Netflix somewhere in London, Maidenhead or Manchester my real world experience might be awful but my local Bristol Speed Test might be great. 

It's only as fast as its weakest link.

Latency over hops - a relay race

The internet very basically works like this. There are series of Routers located in data centres and exchanges all over the world and they talk to each other.  These boxes decide where our data is routed in order to find the end point we require.  The data is divided up into packets and they are addressed. The Router reads the packet's addresses and passes it to another and so on.  These data exchanges are called hops.

We can trace these journeys using simple tools.  Here's an example:

Traceroute has started…

traceroute: Warning: bbc.co.uk has multiple addresses; using 151.101.128.81

traceroute to bbc.co.uk (151.101.128.81), 64 hops max, 72 byte packets

 1  my.house.router.LAN (192.168.x.x)  0.475 ms  0.168 ms  0.153 ms

 2  * * *

 3  aztw-core-2b-xe-333-0.network.virginmedia.net (62.252.231.73)  16.805 ms  10.433 ms  12.965 ms

 4  * * *

 5  eislou2-ic-3-ae0-0.network.virginmedia.net (94.174.238.226)  17.053 ms  18.364 ms  24.068 ms

 6  157.52.127.6 (157.52.127.6)  14.825 ms  14.340 ms  14.059 ms

 7  151.101.128.81 (151.101.128.81)  13.051 ms  14.387 ms  19.881 ms

Above I am tracing the route between my machine and bbc.co.uk and there are seven hops.  We can see the time in milliseconds between each hop, although Routers 2 and 4 are set to ignore this request.  Router 1 is the router in my house.  The packets are handed from machine to machine like a relay race and yes packets can even get dropped and need to be sent again.

The latency between my router and bbc.co.uk is the sum of all of these hops and in this case it's pretty quick.  In some cases, however, we might have more than 40 hops to find a specific site and my connection speed will be restricted by the slowest.  Early on a Sunday morning we will get completely different results to a busier time. 

Real Life speed is not constant

This is why internet speeds vary.  You are competing (contending) with all the other people who are making numerous requests at the same time.  Let's say there was a huge news event like the death of a famous person perhaps or an attack on London (God forbid).  The BBC website will get very busy and so will the Virgin core Routers.  The result is a slow response or sometimes even a failure to get a response and parts of the chain are overwhelmed.

Router 2 in my example interests me.  Every trace I do I get the same response up to router 5, which varies.  So I can assume the first few are local to me.  I suggest that Virgin deliberately don't want me to see some of this as I could measure my local performance.

Truespeed is a bit different

With TrueSpeed what we get is a dedicated fibre all the way to your home which only carries your data, so security and speed is better. It's uncontended so if your neighbour's kids are all streaming high definition video your connection in unchanged.  This fibre connects to a cabinet which is generally local to you in your town or village.  TrueSpeed's core network is connected to this cabinet.  The theoretical throughput of the local cabinet and core network is huge whereas BT's, for example, is shared by hundreds of ISP and huge numbers of customers. In the UK if your using anybody other than Virgin you'll be on BT's infrastructure. That is unless you are on one of a small number of independents such as TrueSpeed.