I want to do a study of the real bandwidth that we get through our ADSL connections in Mauritius. I want to show that we are getting only a subset (I would say < 40%) of what we are actually paying for. I want to do the survey over 7 days (from Monday to Sunday) on at least 5 different ADSL lines (i.e. I need 4 volunteers). I think that we can time how long we need to download a 1Mb file from the US and Europe at different times (say, around 8:30, 16:30 and 22:30) This will obviously produce 5 x 7 x 3 x 2 = 210 different bandwidths which when averaged will produce a pretty good estimate. Who wants to help? I need 4 people with MyT at 256kbit/s. It would be best to start the experiment on Monday 2 July to cater for the 1Gb limitation. As you can easily calculate, the experiment will only use 7 x 3 x 2 x 1 Mb = 42 Mb of your 1Gb quota... There is an ongoing discussion on Facebook on this.
wanna help but the time!!
U know 8.30 normaly we all leave home 4 work n 4.30 on the way back home :)
Rajiv, the times I mentioned are only indicative :-)
But, in fact, the tests should be done mostly at around the same time for the statistics to be coherent.
Incidentally, I am writing a small script to automate the data collection process and this script might even be launched from a cron job or something…
One additional thing I’ve thought out is to time the download of a 1kb, a 10kb, a 100kb and a 1Mb file instead of only one 1Mb file to know what kind of throttling or traffic shaping MT is using…
Hello!
That sounds great! I’d be available… but I don’t fill in the requirements: I’ve a 128 kbits/s line from Telecom Plus…
If at any time you want to extend the types of ADSL lines, you may count me in :)
i’m ready to help but i’m using 128k.
The bandwidth for My.T isn’t that constant as it is for Wanadoo…
& it also depends on the server we’re connecting to…
Why don’t u do different surveys for each type of connection??
I mean one for 128k, the other for My.T 256k & for 512k… :-)
count me in… i am home most of the time these days.
though we need to download from servers which are known to have decent speed, this way, telecom (or any other isp), can’t say “the file was hosted on a machine with slow connection”
why not put a small program to note the down rate a users gets on average then submit the databa on a database somewhere? this way we can be sure the user hasn’t faked the results (am thinking of users who always complain here).
I would be interested to see the results of this experiment. My own rather ad-hoc tests using my 512kbps Wanadoo line showed that on weekends I was able to download from UK Ubuntu repos at around 50kB/s, but during weekdays this fell to around 20 if I recall correctly.
There are so many variables to consider in such an experiment that it would be trivial to formulate an excuse if one wanted to rubbish your results and conclusions. Lets also not ignore the fact that you buy “upto 128/256/512/1024kbps”. Nevertheless, I still think such an experiment is worth conducting.
Good luck with it!
interesting
i don’t fill in the criteria but if you wanna extend to telecom+ 512K
might be able to help
You’re talking about the following:
“It would be best to start the experiment on Monday 2 July to cater for the 1Gb limitation”
1GB limitation?
So you’re cock sure that as long as your 1GB limit has not been passed, you’re guaranteed to receive 256 kbps to the max? And what about afterwards?
No.
In fact, I don’t perceive any difference in performance when under and over my 1Gb limit – it’s always unpredictable.
I’m a proponent of high bandwidths and I’m ready to help. Too bad I don’t qualify. I have a 512K unlim. from T+.
I believe speed obtained has a relation to the IP you get (i.e to what gateway you are on). I saw that 196.192.100.xx is slower than 196.192.106.xx for Wanadoo ADSL at least. Anybody experimented that?
MT JOKE
Ban Russes ti p fer ban fouilles archeologiques et apres 10,000 metres, zot in decouvert ban fibres de cuivre. Ban expert dans Kremlin in tombe a la conclusion ki ti ena system dial-up depi plus ki 1,000 ans avant Jesus Christ.
Lot coter globe, ban Americains ti p fouille trou pou rode petroles (enfin, sa zot ki p dire, personne pas coner ki zot ti p looker dans trou!!), mais apres 20,000 metres fouilles, zot in trouve ban fibres optik. 1 decouverte ki in pousse ban expert dans Nasa a penser ki ti ena system ADSL dans US pres 1,500 ans avant Jesus Chist.
Ban travailleurs Mauritius Telecom ti p fouille trou dans Maurice pou fer passe tuyau et apres 200 metres fouilles zot pas in trouve nanien, 1 decouverte ki in pousse tou ban experts entier pays realiser ki ti ena system Wi-Fi depi avant civilisation commencer!
Nice joke. I didn’t get it first. And then I understood :-)
As for the survey, thanks all for your help. I’m finalising a methodology to follow. I’ll send a mail to some of you today.
I’ve already put some files (a 1Kb, a 10kb, a 100kb and a 1000k) on noulakaz.net which is found in the US and on a free hosting service in France.
Thanks for your help. Let’s start!