• 3 Posts
  • 42 Comments
Joined 1 year ago
cake
Cake day: June 15th, 2023

help-circle





  • It isn’t really about the maximum number of client devices, it’s what they do, what type of standard are they a part of, how far away, the interference (!). This is why it’s pretty much impossible to put a number and say: hey, this TP-Link router will handle 30 client devices, while this Asus router goes up to 100… In a sense, a multi-client stress test kind of addresses this issue, but it kind of doesn’t. It’s because it’s extremely dependent on the conditions that the tester has in their lab/office/home.

    One thing to check on a review could be the attenuation as a better factor than the distance (this way, you can reproduce the result in your house even if just with a single-client test).



  • I understand perfectly where you’re coming from and I would love to find some way to objectively test wireless networking hardware which can be easily replicated in pretty much any other site. The Octoscope tools (now assimilated by Spirent) may be the closest since we get to put the wireless AP/router in a box and then simulate the conditions we want. But, for those that don’t have fat wallets, I guess these open-source tools are good enough, even if the results are heavily subjective. I know that people don’t like good enough, but at this point in time, even the fine edge between realistic and unrealistic is better than nothing.

    And funny thing, WiFi 6 is not really that much better than WiFi 5, unless some very specific conditions are met - I’ve seen it in testing. So yes, I know what hype and advertising can do…

    Gamer Nexus are my favorite when it comes to PC hardware as well and I would love to see them give it a try at testing wireless networking hardware. Who knows, maybe they’ll create a standard for testing these non-enterprise wireless routers.



  • Can your ISP? If so, yes. Because ~25 Mbps * 4 is not a lot of data. And the NAT for four clients mapped to the same firewall/router is pretty trivial. And no, adding “browsing” is not going to be an issue.

    On paper, it is not a lot of data, but then adding more clients requesting 25Mbps continuously and then adding some spontaneous, but intensive web browsing can lead to latency spikes. And the user no longer gets a good streaming/browsing experience. I’ve even seen it on an expensive (by consumer-based networking standard) router such as the GT-AX6000.

    So what matters is the amount of packets per second that can be processed which these speed tests already cover (albeit, somewhat obfuscated because most people don’t understand the network layers.

    I am just trying to better understand this stuff, so I have to ask if seeing how long it takes for a client device to accomplish a certain task wouldn’t be better than just glancing over the average Mbps in a graph? That’s what most publications are showing.


  • Oh, I agree wholeheartedly that collecting the data is not that much fun, especially since yes, I will have to do it. But I think users may benefit to see if the non-enterprise wireless routers can accomplish a certain task. For example, can that expensive Netgear router actually handle four client devices streaming 4K at the same time? What if we add browsing in the mix? The point of this thread was to get an idea if it’s actually worth running these tests (which take quite a bit) and if people are interested in seeing this type of data on the web.




  • Seriously Lemmy is the best. Few minutes in and people are already answering questions. The concept behind the multi-client tests are to SSH into the server and then simulate whatever type of traffic one wants. I have already done it on a couple of wireless routers and it worked great. I hope at least. Iperf can’t really accomplish this, as far as I know, it’s only one instance on a single client at a time. Netburn seems to be the best tool so far, while keeping things free and open-source.