Why You Shouldn’t Rely On Speedtest.net
Who doesn’t like free internet? Specially when it’s a WiFi with a good signal strength. If you’re tech freak like me, the first thing you’d probably do is go to speedtest.net and checkout the download/upload speeds. Well, without wasting much time, I’m here to tell you that those results aren’t true, atleast when you’re having high-speed access.
While upto 100MB/s connections might show up to be accurate on their website, connections above 100MB/s are pretty doubtful. I have done some research and I will propose 2 reasons for the mismatch and inaccuracy of these speed results.
1. High Latency/Ping
2. High Traffic/Load on Speedtest.net Servers
Speedtest uses multiple servers in a location to run speedtests for it’s users and as far as I have observed, there are only few locations that support a 10 gigabit port. For those who don’t know, a 10 gigabit port means you can max out the internet speed to 10 Gbps which is pretty much rare.
Now I happen to own a dedicated server that is hosted on a 10 gigabit port and as soon as I have access to it, I open up the speedtest website and run a test.
Results:

Pretty annoyed, I go to my server provider and ask them to fix it. Before I receive their reply, here is another result:

This gets me thinking they have started finding a solution to the problem already and I get a little grin on my face.
Soon enough I receive their reply and they tell me they have started the process which means the improvement in the result above wasn’t due to them fixing stuff. I get curious, but I nod and give them a green signal to proceed with it.
Next Result:

I was expecting atleast 6-8Gbps when 10Gbps was promised to me. Fed up of all this, I initiate a refund request when suddenly I remember I had to download and backup something on the server. I setup Internet Download Manager (IDM) on the server for the download and guess what? I hit 80MB/s for a single file which got me putting a few more downloads simultaneously and holy crap, I was reaching over 1.2Gbps easily.
Please note that MB/s > Mbps.
1MB/s = 8Mbps
Now 1.2Gb/s wasn’t what I was looking for, but it was far better than the 300Mb/s I was being showed up on the speedtest results which clearly proves those results inefficient.
The theory behind all this is that the downloads from a single link/server are capped upto a few MB/s. This is not an error or some server incapability, but a cap limit put up by the server companies and ISP’s to prevent excessive usage of their bandwidth. When you put up simultaneous downloads, you reach different servers and try to pull in maximum data from them which gives you a boost in the download if measure altogether. Torrents on the other side run on UDP ports unlike IDM which uses TCP ports, and since UDP connections guarantee better speed, running torrents simultaneously with IDM running multiple downloads already will give you the actual speed result.
Here is a direct link to a Windows 7 Copy from Microsoft which I used in the above tests to measure the speed.
http://msft.digitalrivercontent.net/win/X17-24209.iso
EDIT: I just some torrent downloads along with IDM, and I was able to reach 4Gb/s. Pretty satisfied now.



