How fast is your net speed?

  • Thread starter Thread starter Yasogami
  • Start date Start date
At my local library:
http://www.speedtest.net/my-result/4331423100

4331423100.png
 
4353046491.png


My wifi connection drops every once in a while, but this speed is tolerable for me.
 
Finally on Fibre-To-The-Premises (FTTP):

# speedtest-cli
Retrieving speedtest.net configuration...
Testing from Launtel (87.121.94.153)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by Host Universal (Adelaide) [17.16 km]: 2.215 ms
Testing download speed..........................................................
......................
Download: 526.12 Mbit/s
Testing upload speed............................................................
..........................................
Upload: 46.56 Mbit/s
 
Kugayama said:
Interesting how it changes over the years and for different people.
Click to expand...

Scenario 1: Traffic from your apartment to the outside world
Speeds might be different on numerous devices on the same router. Reson is that the network "cards" are differently bond into each machine.
Imagine you have two PCs with different hardware in the same network, being routed over the same router.
These two might have different network speeds.
If you have two equal devices on the same network being routed over the same router, speeds might be equal, but can then depend on factors outside of your household.

Which there are... (not a complete list, just examples)
- bandwidth and latency from your household towards the swtching center close to you
- bandwidth and latency from the switching center towards their computing centres
- (if traffic goes to another country) bandwith and latency from their switching center towards the exchange point <-- mostly overloaded
- (if traffic goes to another country) bandwith and latency from the exchange point to the network in the other country <-- mostly overloaded
and so on and so on..

There are so many factors. Traffic speed is never the same, because there are too many factors that come in your way for a dependable measurement.
That's why even same source towards different measurement services... will bring other results. (not only but also because these services also have their bandwidth.. and if too many people are currently testing, results may differ for your measurement), and so on.
So.. in short.. these measurements "trusted network into a different network" can only be an indication, but never a dependable measurement.

Scenario 2: Traffic from PC1 to PC2 or PC3 in your own apartment over your own local network
Best (dependable) results will be achieved between network bandwidth measurement devices (possible FLUKE devices). These things come with extremely good network hardware. They also do not transfer data in terms of "human readable" data. These devices generate sample data in form of network packets that should fit exactly (stacked) into your network bandwidth at home. (max it out and see how many packages in size X can be transferred and recieved at the same time)
And yes, you always need two of exactly these... one sender and one reciever.
With a PC or notebook this is different, as their network transfer works a different way. These packets are packed up in the network driver and then transferred over the hardware network "card" into your network... where there are also are different reciever network components, that unpack these packets, compute them and repack them for forward routing to the target device in your network.... where it will be recieved, unpacked, and computed.
This takes WAY longer and has lots of delays ... produced by your hard- and software environment.
Even between three equal PCs or notebooks results will differ. (too many factors still left.. like driver settings, hardware quality, network cable quality, switch/hub hardware quality, and so on and so on)
So.. even in your local network these measurements are not dependable and will change.


We once had this demonstrated in the company. We bought a steady internet connection of 1 Gbit synchronous (up and down), but our tests only brought outcomes of max 750 MBit/s down (numerous different results with numerous different target test services). And yes.. it's very hard to find testing services allowing 1 GBit traffic. Most of them only allow way lower bandwidth per connection. My boss wanted to see results like 1 GB per second. 750 MBit is nowhere near 1 GBit... so we called them up.

The telecommunications company brought their local expert (and I mean a NERD... he came in sandals in winter and knew the whole lot about every network layer etcetc).
He had these FLUKE devices with him, and he did run numerous tests with our hardware to show us that our measurement methods are inappropriate.
We then tried his FLUKE devices in our local network. - Target speed also was 1 GBit. While our own hardware also could never reach this, his FLUKE devices could.... because these are powerhouses and have NICs of very high quality. (that's why these things cots the equivalent of a new medium size car... each!). They do not build these network packages, they construct them by parameters and the other one recieves them in the same fashion the first one sent them.

With these new findings we agreed that we could run a test... our network towards their local switching centre here in the city.
For this one of his FLUKE devices went to this switching center, where they also had one testing machine at a certain IP.
As soon as this test was set up, we got a call and then he brought the 2nd FLUKE device with him (again sandals in winter) to connect this on our side.
Result was: Our internet connection was even a small bit faster than expected.. 1 GBit and some extra bits up and down.

So.. in the end.. all these internet speed tests are FAR from precise. You should rather not do them, as they fill up valuable bandwidth and cause delays for other "normal" users on and across the network(s).

To find out what's going on, you need to convince such a tech nerd with ultra expensive FLUKE network testers to measure your bandwidth.. ; )

🔴 Cif ~
 
You must log in or register to reply here.
Back
Top