Those of us who complain about slow Wi-Fi are often left scrambling for answers.And scramble, I have.I tested 20 different ways to improve my Wi-Fi network, but to be able to tell which are actually helping, I had to devise a reliable plan for testing speeds.
Local iPerf testing tells me what the network itself can actually do Remove the internet from the test completely Close If you want to know the actual speed of your connection, the first thing to do is find the right tool.I used iPerf3, a network-testing tool that measures throughput directly between two devices on your local network.In that setup, one device acts as the server, the other as the client, and the test traffic doesn't have to leave your home.
That gives you a clearer picture, as it doesn't rely on any third-party servers, websites, or apps; this is all about your home connection.The ideal setting for this is to use iPerf3 on a PC connected to the router or switch over Ethernet.If you follow my coverage here at How-To Geek, you've probably gathered just how big a fan of Ethernet I am.
In fact, moving some stuff off of Wi-Fi massively improved my home internet.In this case, doing this gives you the best stability, and it shows you a baseline for what the LAN itself can handle without Wi-Fi or your ISP muddying the results.And the test itself will quickly reveal if there's something wrong, because if your network is supposed to be 1Gbps but iPerf is only pushing a fraction of that between two wired machines, you'll know there's a local bottleneck hiding in there somewhere.
Next step? Run this same test on a Wi-Fi device connecting to that wired server, using its local IP address.That's a test for the wireless link that also keeps the ISP totally out of it, which, if you ask me, is great news—I've not had the best luck with ISPs.Related 6 networking upgrades that are a giant waste of money for most homes I’m a tech enthusiast, and I’ll never buy these 6 expensive networking devices again Posts 15 By Monica J.
White A LAN file copy shows whether fast benchmarks translate to real work The disks and SMB get a vote, too My testing adventures don't stop here.iPerf3 is good because it isolates the network itself, but I also like to copy one large file between two devices on the LAN.That gives me an idea of what the connection feels like when I'm actually doing something, which, to me, is ultimately a much better way to measure than just relying on synthetic benchmarks.
"A big file" isn't a random suggestion, by the way.That's a better option than a folder full of tiny files, because small files add extra filesystem and SMB overhead.That can make the transfer look slower for reasons that aren't related to the network itself, so it kind of defeats the purpose.
Be careful not to run into a different sort of bottleneck, though.The storage on both ends of this little experiment needs to be fast enough to keep up.If iPerf3 gets close to saturating the link, but the files are crawling along instead of copying at a decent pace, don't immediately jump to the network as the culprit: check the drives, SMB behavior, CPU load, and background processes first.
A wired speed test is the best way to judge your connection Now I test beyond the router At the risk of repeating myself, I'll say it again: a wired test is the best way to do this.So, once I know that the local network is doing A-OK, I bring the actual internet back into the picture.For this part, I recommend connecting a PC to the router over Ethernet, then turning off Wi-Fi entirely.
Then, run a regular internet speed test to see what your ISP connection can deliver without being bogged down by things like signal strength, interference, or distance from the router, as well as all that other Wi-Fi-adjacent nonsense.This is also where your router's own speed test comes in as a useful point of comparison.Is it better? Is it worse? If it's fairly close to your plan's full speed, the WAN side is most likely doing a good enough job.
But if the router looks great and the wired PC is much slower, that means issues that aren't router-related, such as the LAN port, the Ethernet cable, the network adapter, or any other number of local bottlenecks between the router and the device.Wi-Fi has to be tested where you actually use it One result beside the router tells me almost nothing We're circling back to Wi-Fi for this final check, and yes, you should only proceed if the wired side has proven itself to be in top shape.That lets you judge Wi-Fi as a standalone connection in your home, and that can save some trouble, as it isolates potential problems and narrows them down to just the wireless side of things.
Coincidentally, this is where a lot of testing goes wrong, because Wi-Fi introduces an inherent element of unpredictability.It has a lot to do with your distance to the router, because getting a good result right next to the device tells you nothing about how well it'll perform once you're a room or two away.For this test to have any meaning whatsoever, I recommend testing a few spots around the home, preferably using the same laptop or phone each time.
At each location, compare both the local and the internet-side results, and that means running iPerf3 against the same wired server, then following it up with a regular internet speed test.If a particular location in your home shows a huge drop in the connection while everything else looks fine, you might be dealing with dead zones.The useful number is the one that isolates the problem Speaking as a person who's spent more time fiddling with her internet connection than I can truly recommend, I'll tell you that there's no magic threshold where your internet speed goes from "slow" to "fast." It's more about what's good enough to you and your needs, as well as what exact number you're looking at.
The more thorough your testing, the easier it'll be to isolate the bottleneck (and hopefully fix it, too).Related I stopped rebooting my router to fix Wi-Fi, and these 7 tricks worked instead Rebooting your router won't fix your Wi-Fi: 7 tricks that actually work Posts 2 By Monica J.White
Read More
