Speed is one of the most important aspects for internet users and while using a VPN can make your connection slower (due to the encryption applied), there are services that reduce the impact, making the difference almost unnoticeable. In order to test the speed of a VPN connection, most people rely on speedtest.net and while this popular website offers good results when it comes to regular internet speed, it may not be the best option to measure a VPN. This is particularly true if the protocol used is OpenVPN, which is the best and most secure option available.
When you use speedtest.net to monitor the speed of your VPN connection, you may get inaccurate results. This is because the website uses a Flash based tool that doesn’t take into consideration LZO compression, which is built into OpenVPN. We will discuss this further throughout the article, but the main thing to keep in mind by now is that in order to measure data transfer speeds over OpenVPN accurately, speedtest.net should be replaced with a more effective method.
What factors have an impact on connection speeds?
The speed of your ISP is only relevant between you and the ISP server
The broadband speed that your ISP claims to support only matters when it comes to the connection between your computer and the Internet Service Provider’s nearest server. Beyond that, data transfer between ISPs and countries can’t be controlled by the ISP and it is a free area. This is not something that ISPs would openly admit but it is a result of the way in which internet works. Plus, it is a factor that explains the different results that you can obtain when measuring speeds. All that you can do is to carry out many tests and get an average of the results.
OpenVPN was designed with the possibility to opt for using the LZO lossless compression library. This is similar to the popular .zip format and compresses the size of certain file types. Although this can increase data, it wouldn’t have an impact on your actual bandwidth usage. If a file has been compressed already and it is is in a format such as .rar, .zip. .mp3 or .jpg, compressing it further wouldn’t make a big difference. As we mentioned at the beginning, LZO compression is a factor that is not taken into account by speedtest.net, which leads to inaccurate results.
TCP and UDP
TCP stands for Transmission Control Protocol and UDP is User Datagram Protocol. OpenVPN can be run over both of them. The difference is that TCP uses error connection to confirm if a data packet has arrived. If no confirmation is received, the packet would be sent again. If confirmation is received, then the next packet is sent. UDP doesn’t verify if the packet reached its destination and it would continue sending packets, even if the previous ones haven’t arrived. This makes it faster, but less reliable than UDP.
Most ISPs apply this cap to a certain extent, which is also known as bandwidth limiting and traffic shaping. BitTorrent traffic is generally throttled but it may also be shaped in a less obvious way depending on peak usage times and other factors. M-Labs offers a transparency measurement tool that allows you to find out how much throttling is imposed by your ISP.
Programs that hog bandwidth
Keep an eye on programs and processes running on the background when you are measuring speeds as they will affect the results. These programs can slow down your connection and hog your bandwidth. This is something to consider particularly when it comes to uploads because the majority of ISPs apply more strict limits on upload bandwidth than on downloads. In general, using over 1/4 of your upload bandwidth will have a significant impact on the speed of your downloads. BitTorrent clients are usually responsible for hogging upload bandwidth, but online storage services can also cause issues. It is important to keep an eye on them and pause them while you are carrying out a speed test.
VPNs affect the speed of your connection because the data has to travel a longer distance between your computer and the internet as it has to go through a VPN server. The farther you are from the VPN server, the slower your connection would be so it is advisable to opt for a server that is closer to your geographical location. It is also important to keep in mind that encryption and decryption will require processing power on your computer, affecting speed as well. Encryption wouldn’t have an impact on the tests and shouldn’t be an issue for modern computers, although it may affect less advanced devices. Keep in mind that the processing power required increases when a higher level of encryption is applied.
Another crucial factor to note is that a VPN server also requires a great deal of processing power in order to handle tasks. If there is not enough processing power on the VPN server, the connection will slow down significantly. Generally speaking, when you connect to a VPN server that is close to your location and you are using OpenVPN, you can experience a speed loss of up to 10 percent compared to the performance of a connection without a VPN. However, this is not a standard result and that figure may vary.
VPN speed testing methods
The results will also depend on the type of broadband connection and VPN provider that you use. To test the speed of your connection, it is advisable to perform multiple tests, first without VPN and then running the VPN. We have carried out three tests for each method without VPN USING 20K band and then followed with three with VPN (using TorGuard). Ideally, tests should be carried simultaneously, but this is not always an option. Our tests were carried after each other as soon as possible to make them as consistent as possible.
When using speedtest.net, the average speed results without VPN were 20.2Mb/s and with VPN, the speed actually increased to 22.9Mb/s. Upload speeds came at 1.4Mb/s without VPN and at 1.2Mb/s with VPN. The issue with speedtest.net is that download results are considerably higher than the results without using VPN. Considering that the ISP used imposed a speed cap, the results are not likely to be accurate and this is due to the fact that LZO compression is not accounted for, as previously mentioned.
Speed.me is an HTML5 web-based test that automatically selects the test server focusing on stability instead of distance. For the server it selected for us, the average download speed without internet was 18.1Mb/s and with VPN 8.1Mb/s. The average upload speed was 1.1Mb/s without VPN and 1.8Mb/s with VPN. The results obtained from Speedof.me show poor download speeds but the main thing to take away from them is their overall consistency.
Testmy.net also uses HTML5 for speed tests in an effort to compensate for compression and offer more accurate results. The average download speeds obtained without VPN were 16.3Mb/s and with VPN (server in our location) 16.3Mb/s as well. However, the results were affected when we connected to a server in a different location (although not too far from ours) and in that case, the speed was just 13.6Mb/s.
The average upload speed without VPN was 1.2Mb/s ( Know what a Mb is ) with a VPN (in our location) 1.08Mb/s. A VPN in a different location gave us 1.07Mb/s. The results obtained with Testmy.net were quite inconsistent, but this may be explained by the fact that the test server that was not in our location, was far away. The discrepancy in the results can also be caused by the variable nature of internet.
While the other testing options in the list are web-based, ttbMeter is a downloadable tool that is offered by thinkbroadband, a independent website that is focused on broadband news and information. ttbMeter offers many configuration options and tests. Unfortunately some of the tests options are not available so we had to stick to a standard download test. The result without VPN was 16.1Mb/s and with VPN 17.3. Although there is some discrepancy, this can be explained due to different factors including the variable way in which internet works. However, the main issue with ttbMeter is that it is not practical as it only shows results for a few seconds and it is not easy to use.
If you prefer a simple, basic testing solution, you can opt for the manual method. In order to carry a manual test, you simply need to download a test file and use a stopwatch to find out how long it took to download. Make sure that you carry these tests using compressed files in order to compensate for LZO compression. Using a 100MB (800Mb) test zip file the average result (of three tests) was 12.4Mb/s without VPN. With VPN, the average was 9.6Mb/s.
To test upload speeds, we uploaded a 5MB(40Mb) file to an online storage service and the average without VPN was 1.07Mb/s. With VPN, the average result was 0.90. The manual method comes with some hassle and it is not the most precise option to test the speed of your connection. However, it provides a realistic idea of how internet speeds are affected by a VPN.
Looking at all the results obtained, you can see that the main issue is inconsistency, not only between different testing methods but also when identical tests are carried with the same tool. This is why it is important to keep in mind that speed test results should not be taken as a norm. While they can give you a general idea of the performance of a VPN, the results would vary depending in many different factors. Consistent testing over time would be the only way to get results that are more accurate.