Как мне точно измерить скорость интернета?

I trust in speedtest.net for a decent idea. I often run it trough different connections throughout the house to validate the numbers. It's been a safe to-go.

Your actual speed, in terms of Mb/s can only be measured when it's under stress, i.e., the test is running right now.

If you ran a constant test, you would
a) have no bandwidth left to do anything else &
b) really, really annoy your ISP.

Sites that I've used for such testing in the past include the following:

1. BandwidthPlace
2. CNET Internet Speed Test
3. Digital Landing Internet Speed Test
4. Speakeasy - now MegaPath
5. Speedtest

You shouldn't expect to get exactly the same results between speed test providers, since the number of network hops between your systems and their servers will vary, i.e., there will be a varying number of routers between your systems and their systems depending on the network path between your systems and their systems. E.g., as an example of different results, you can see the bandwidth test results I found several years ago when testing a Verizon network service using the above test services, including tests run using the same test provider, but at different times or with servers in different cities for a bandwidth test provider.

You can see the number of intervening routers, i.e., "network hops", between two systems by using the traceroute command; on Microsoft Windows systems it is tracert, e.g., tracert www.speakeasy.net. You may notice that the number of network hops varies even when running a test with the same provider from one day to the next, since the way networking works is that if one network path becomes unavailable or congested, another, if available, may be taken, just as you might choose a different route to drive from point A to point B if you know there is a congestion on a particular road between the two points.

You are unlikely to get exactly the same results even if you run several tests using the same Internet speed test provider, since there will be other traffic between you and the provider's test server and that traffic will vary from moment to moment. If there is more congestion on the path, your speed test results may be lower than when you previously ran a test.

As far as running a program on your computer to judge the available speed to a system outside of your local area network (LAN), Iperf is frequently used to test the available bandwidth between two systems. It is command line program available for Microsoft Windows, Apple OS X, Linux systems, as well as iPhones and Android phones - see the download links at Iperf.fr. To use the program to test the available bandwidth between two locations, you will need to install it on two systems at those locations - perhaps you could install it on a system at a friend's or relative's house if you don't have two systems of your own at separate locations or you could test from a phone or laptop to another system at your home location while outside of your home, if that is the location whose bandwidth you wish to test. You would need to configure your home router/firewall to allow the incoming connections and also might need to configure firewall software on the host system at home, if you are using host-based firewall software, such as Microsoft Windows firewall software. When using Iperf, one system is run in "server" mode and the other in "client" mode. You can find online tutorials on its use.

There are also public Iperf servers you can use to test your bandwidth to various locations throughout the world - see the list at Public iPerf3 servers. E.g., you could install the Iperf software on one of your systems and run the command iperf -c ping.online.net - the -c means your system would be acting as the client in a client-server relationship. Here is sample output from a test I ran showing the maximum speed achieved for the test was about 8 Mbs from the particular system I used for testing:

$ iperf -c ping.online.net ------------------------------------------------------------ Client connecting to ping.online.net, TCP port 5001 TCP window size: 22.5 KByte (default) ------------------------------------------------------------ [ 3] local 192.168.0.5 port 58224 connected with 62.210.18.40 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-10.1 sec 9.38 MBytes 7.78 Mbits/sec 

I find that java based speed tests like Ookla's speedtest.net slows down the system, and when working on other people's computers they might have outdated or non functioning java etc.

I quite like speedof.me and dslreports.com. Both of these are html5, so you can do the test on a mobile browser too instead of installing an app.

But yeah, speeds do vary at different times of the day, the week, peak times around holidays etc.

Another thing that adds to the complication and skews speed tests are the so called "bursts" that some ISPs give downloads for the first 20 MB (or other variable amount), hence the need to test under stress or longer duration. There's more to it than that, the tests themselves seem to be overly flattering of the ISPs supposed speeds.

I find it more telling to observe what a connection's capability is to see what the speeds of a massive sized popular bit torrent will download at. Not accurate or "scientific" like you're asking. But it's a pretty real-world guage.

Another tool to consider (if it's speed you're after) is namebench, the DNS you use makes a difference in speeds (and security) as well.

I've worked in technical support for a telecommunication company and we would do 3 speed tests with customers and call the average of those 3 tests their speed. It might be worthwhile to ask benchmarking experts though like at pcmag.com or smallnetbuilder

If you want to track your internet speed for the entire duration. I'd recommend "Netspeedmonitor". It is an open source application which runs in your taskbar constantly tracks your internet accurately.

If you like to try this application, then click here