the bigger question is why do you need to test the line every 3 hours?

do what most others do, test it when it seems slow

Talkiet:

 

There are some reliable ways, but most of them require either having access to a server connected to your ISPs core that YOU control and are the only user of - although for only 200Mbit that may not be critical. Wget-ing a large file from a known server is actually a pretty good way to measure TCP single thread throughput, and iperf is good as well (BEWARE version and OS interoperability performance issues!).

 

As an example of what you CAN do when you have access to put things on the network in order to test (for example) a new high speed fibre service that many people (not us) have called gigabit, is this... (Forgive the extreme sanitising!)

 

 

Cheers - N

 

So in principle one could automate the downloading of a decent sized file with wget from say GoogleDrive and pipe the screen output to a file, extract the relevant numbers and use that and then do the same for an upload?  I see that some of the speed measurement tools seem to use much smaller file sizes than 4GB too.

SteveWright:

 

Talkiet:

 

There are some reliable ways, but most of them require either having access to a server connected to your ISPs core that YOU control and are the only user of - although for only 200Mbit that may not be critical. Wget-ing a large file from a known server is actually a pretty good way to measure TCP single thread throughput, and iperf is good as well (BEWARE version and OS interoperability performance issues!).

 

As an example of what you CAN do when you have access to put things on the network in order to test (for example) a new high speed fibre service that many people (not us) have called gigabit, is this... (Forgive the extreme sanitising!)

 

 

Cheers - N

 

 

So in principle one could automate the downloading of a decent sized file with wget from say GoogleDrive and pipe the screen output to a file, extract the relevant numbers and use that and then do the same for an upload?  I see that some of the speed measurement tools seem to use much smaller file sizes than 4GB too.

 

Yes, you could. But don't do it. It's pointless and it wastes resources. If you ARE going to do it, you'll get reasonable results if you can find a web host under 20ms away and using a 5MB file. Once an hour would be EASILY often enough to detect any user impacting issues.

But again, it's not worth it. If you want to do something as a learning exercise however, totally reasonable project. Maybe try it with latency or jitter. You'll pick up near congestion using a jitter measurement before it becomes noticable in throughput testing.

Cheers - N

Jase2985:

 

the bigger question is why do you need to test the line every 3 hours?

 

 

 

I don't understand this either. Raw speed is pretty meaningless when it comes to determining the quality of a connection.

QOE is all that matters - and a connection delivering you your 200Mbps could easily deliver a very poor real world experience. If you want to be testing a connection for "quality / performance" then continually running a speedtest is not going to give any meaningful results. Testing for latency or jitter on the other hand tell a lot about the quality of a connection and how it will perform in the real world.

Running a speedtest every hour is a bit like looking at the motorway webcam, seeing there is no traffic and declaring that the motorway is working perfectly. Little beknown to you the road surface is full of potholes which is actually killing everybody's suspension and tyres.

Talkiet:

 

Yes, you could. But don't do it. It's pointless and it wastes resources. If you ARE going to do it, you'll get reasonable results if you can find a web host under 20ms away and using a 5MB file. Once an hour would be EASILY often enough to detect any user impacting issues.

 

 

 

But again, it's not worth it. If you want to do something as a learning exercise however, totally reasonable project. Maybe try it with latency or jitter. You'll pick up near congestion using a jitter measurement before it becomes noticable in throughput testing.

 

Cheers - N

 

 

Thanks for your help!! Very much appreciated.

I will probably just use what I have at the moment but now have a much better understanding of the results that Ookla provide.  If I get some really slow results I'll run a wget to get a better idea of what I'm actually getting.  The other good thing is that I can get a feel for actual speed any time I download something on my Linux box with wget.

sbiddle:

 

Jase2985:

 

the bigger question is why do you need to test the line every 3 hours?

 

 

 

 

I don't understand this either. Raw speed is pretty meaningless when it comes to determining the quality of a connection.

 

QOE is all that matters - and a connection delivering you your 200Mbps could easily deliver a very poor real world experience. If you want to be testing a connection for "quality / performance" then continually running a speedtest is not going to give any meaningful results. Testing for latency or jitter on the other hand tell a lot about the quality of a connection and how it will perform in the real world.

 

Running a speedtest every hour is a bit like looking at the motorway webcam, seeing there is no traffic and declaring that the motorway is working perfectly. Little beknown to you the road surface is full of potholes which is actually killing everybody's suspension and tyres.

 

 

Yeah I'm getting it now.

do you work from home or does it run 24/7 when you are at work and asleep?

Jase2985:

 

do you work from home or does it run 24/7 when you are at work and asleep?

 

It runs on a home server as a cron job every three hours and I have been using the results to keep an eye on Vodafone's network consistency - sounds like I'm using the wrong solution based on the above but the results have been fairly consistent to date and I am reasonably happy that I'm receiving about what I'm paying for though time will tell with my new 200Mbit service.  Most of this is just a hobby for me but I'm learning a fair bit, fairly quickly .  I'm learning to code perl, bash, html, php, css and anything else that I need on the fly.

@SteveWright

Have sent you a PM re speed testing on cable.