Internet speeds on new provider

raykloss · May 18, 2021, 8:15pm

So I recently upgraded my Internet from U-verse to Spectrum in the hopes of getting more speed to my entertainment center and my general Wi-Fi. I have a Linksys router and mesh network and when I test the speed at the Linksys, it shows a very high speed of download, but none of my other speed tests via Wi-Fi show much increase in speed.

My iMac Pro is connected to the WiFi and is 8 feet away from the router/WiFi, but shows a speed of 32 when the Linksys app shows 400. At the router. I have an Ethernet cable coming from the router to a switch, to split the Ethernet, but am considering wiring it from the Linksys directly to the Mac (which I did and a direct ethernet from my router, I get 45 DL and 11.4 UL). The switch is a Netgear GS108 which should be gigabit and I think my cables are all 5e or better. On the other hand, when I connect my iMac directly from the modem, I get full speed of 400.

Now I am the first to tell you I know next to nothing about networking, though I have worked with Macs since the beginning. I am trying to figure how to optimize things so I am getting what I am paying for, which should be 200 DL. All new Ethernet 6 cables from the modem to the router and back to the switch have not affected the speed. I am thinking it is the switch, but as a Gigabit switch, shouldn’t it be able to handle the speeds? Do switches go bad?

raykloss · May 19, 2021, 4:05pm

The short question is: Do Ethernet Switches go bad or not pass through the same speed for some reason?

Shamino · May 19, 2021, 4:50pm

They don’t “go bad” in the sense that there is a limited lifespan, but all electronics can fail, whether due to a power surge or overheating or liquid damage or a faulty component.

Your GS108 is gigabit. It shouldn’t be slowing down any data (unless there is congestion from other attached devices). You may want to check to make sure the links are running at gigabit speeds. The link lights above each port should tell you. According to its datasheet, on page 2, the right LED means 10M operation, the left LED means 100M and both together means gigabit.

If one of your links is not gigabit, check the other end of the connection. Make sure it is a gigabit-capable device and that it is configured for gigabit operation. Your Mac should have gigabit capability, but there are settings that can limit it - make sure the settings are set for automatic configuration or that they are manually configured for gigabit (1000baseT):

If the link to the Mac is running at Gigabit speeds, but the link to the router is not, make sure the router’s LAN port is actually gigabit capable. Some older (or cheap ones sold today), only have 100M LAN ports.

Similarly, different Wi-Fi technologies have different speed limits. If your router is older and only supports (for example) 802.11g, it’s going to have a theoretical maximum of 54 Mbps. If it is 802.11n, that maximum speed can be anywhere between 72 and 600 Mbps, depending on the router’s hardware capabilities, the connected device’s capabilities and how each are configured.

What is the router’s model number? This will help us determine what it is theoretically capable of, which is important to know before trying to troubleshoot everything else.

raykloss · May 19, 2021, 5:55pm

Thanks for the answers and questions.

The router is a Linksys Velop Intelligent Mesh WiFi System, Tri-Band, 3-Pack White (AC6600) which is gigabit ready. The switch shows the ports that come from the router, go to my iMac and go to my Mac mini have both lights blinking. (Other Ethernet ports are to a ups battery - no active signal - and to a canon printer - one light blinking). The Network configuration looks ok also.

When I test the internet speed at the computer, it is 47 Mbps and 33 on WiFi. When you check the speed at the router (with a Linksys iOS app) shows input at 482 Mbps and uplink at 12.

I just checked the speed at my iPad Pro and it reported 185 mbps on WiFi. None of this makes sense to me.

Shamino · May 19, 2021, 6:34pm

Thanks. So it looks like we don’t have a hardware issue. It looks like there may be a software issue. Are you seeing an actual slowdown during real work (e.g. when downloading large files) or only when running speed tests?

It may be that the speed test tool you’re using isn’t very good - which may easily be the case if it’s a browser-based tool.

If you have some kind of firewall or network performance-enhancing utility running, it may be giving low priority to speed-test packets in order to give real traffic higher priority.

Is the other computer sharing the Ethernet switch (the Mac mini) isn’t in the middle of a large download (e.g. a background system update)? That could easily consume the bandwidth your speed test wants to use. Does anything change if it is disconnected from the network?

Other computers wouldn’t affect Linksys’s own tool, because those tools typically run on the router itself and may even cut off the rest of your LAN during the test, in order to prevent interference.

Another interesting thing to look at (although I doubt it will solve anything) is the nature of your Wi-Fi connection. If you hold Option when clicking the Wi-Fi icon in the menu-bar, you will see detailed status of your current connection:

Lines of interest include the PHY mode, the channel bandwidth and the Tx Rate. If you’re seeing something that looks wrong (e.g. a PHY mode of 802.11g when your hardware supports 802.11ac), there may be something misconfigured.

But that would have no effect on a wired Ethernet connection.

raykloss · May 19, 2021, 7:28pm

I was using the Ookla Speedtest on the iPad and the iMac. I think it is pretty reliable. I verified the iMac speed with a test in TechTool and the results are pretty much the same. The Linksys has it’s own test that gave me the 400 speed.

No software is running or downloading anything.

I still don’t get why the iPad 9.7 gives me fast speeds (but my 12 iPad Pro is the slower 37). I just want to get my money’s worth on the Spectrum, but will definitely downgrade from the 400 mbps special to the 200 regular.

My Apple TV 4K gives the 38 mbps speed, but I think/hope that is good enough for 4K.

Shamino · May 19, 2021, 8:50pm

Ookla works by sending and receiving traffic with a nearby server. It doesn’t measure your link speed, but the overall throughput between yourself and the specified server.

You should be able to get (within the app) a list of servers that are near your location. It is critical that you pick one that is close to you (ideally hosted by your ISP), because otherwise you end up measuring all of the network nodes in between, which will artificially reduce the measured bandwidth.

Also, it may be the case that a single computer simply can’t keep up with a high speed like 400 Mbps. This will especially be the case for mobile devices, but may also be the case for desktop/laptop systems.

Since you have two Macs on your LAN (iMac and mini), you might want to enable file sharing and copy a file from one computer to the other and see how long it takes. I can guarantee you that you will not see actual gigabit speeds, but the speeds you do see may be useful as a basis for comparison.

(A better baseline would be to run a pair of speed-test apps, one on each computer, in order to eliminate the overhead of the SMB network file system protocol, which isn’t particularly fast. But I don’t know of any specific apps for this).

Similarly, you can try downloading a large file from a web site and see how long it takes. For instance, the current Firefox installer is 134 MB.

Regarding the Linksys test, I’m not sure what it is actually measuring. It might simply be due to the fact that its network processor is designed for high speed connections and therefore can keep up with a high speed remote benchmark server.

Ultimately, you may find that no single computer in your home can actually keep up with a 200 Mbps link, but that may not be a problem if you have many devices that all try to access the network at once. For example, in my home, I may be conducting a Zoom call on one computer while another is downloading a software update, while my wife is streaming Netflix while my daughter is installing a game on her XBox.

ashley · May 19, 2021, 11:42pm

I’ve used iperf for this many years ago. You run an iperf server on one machine, and measure the speed to it from an iperf client on another. It’s a command line thing, but a relatively straight forward one, so if you are comfortable with the command line, it should be ok.

Two articles about it (although I can’t say whether they are accurate - I last did this in 2012): https://www.macobserver.com/tmo/answers/how-to-test-speed-home-network-iperf and
Installing iPerf on a Mac OS X system.

schwartz · May 19, 2021, 11:59pm

fast.com tests with Netflix servers, who actually pay for priority access. Best for finding absolute highest possible rate.

Simon · May 20, 2021, 12:55am

Excellent advice. I like using fast.com too. It’s minimal effort for a result that’s usually close enough for a good first estimate.

morwen · May 22, 2021, 11:21pm

Didn’t you say you had a switch between your router and your Mac and when you take that out of the picture your Mac gets higher speeds?

Shamino · May 23, 2021, 5:02am

The exact quote was:

Which means bypassing the router as well as the switch, which is interesting.

It is more likely that the router is causing a slowdown than an unmanaged Ethernet switch (which should have full non-blocking capability on all ports if it isn’t complete junk - and Netgear is a good brand).

His router is a Linksys Velop AC6600 set. Unfortunately, I couldn’t find specifications mentioning its maximum throughput when routing traffic from a WAN port to a LAN port. Although it has Gigabit Ethernet ports, the forwarding chip, when not in bridge mode, might have a limit that is lower than gigabit speeds.

I also don’t know how his local topology is set up. If the Ethernet switch is connected to the same node that connects to the modem, or if it is connected to some other node in the mesh. And whether the mesh nodes are connected to each other using Wi-Fi or Ethernet.

Right now, there are still too many unknowns to be able to conclusively determine what’s going on.

nick3 · May 23, 2021, 10:10pm

Just some speculation, but the 400 figure that is quoted looks like the speed of the internet connection from the comment about planning to downgrade from 400 to 200. When the OP mentions that the Linksys router reports that speed I think what may be reported is the sync speed of the connection. It would be very unlikely that one would see exactly 400 Mbps across the connection to the internet from a computer although you might get quite close to it.

Is it possible that there is some confusion here between Mbps and MBps? 47MBps is 376 Mbps, which is just under the sync speed. One of the slightly suggestive points is that the OP sometimes mentions the number without any qualification.

Just a thought…

raykloss · May 23, 2021, 11:18pm

The Linksys iOS program is supposedly measuring the speed to the main device I am using as the router.

jbgurman · May 25, 2021, 1:40pm

Sometimes individual ports in more expensive, enterprise switches fail. In consumer devices such as the GS108, it’s more normal to find the ports served by only two chips (that is, four ports on each). When I had a lightning strike very close to my house some years back, my GS608 lost four ports.

Iirc, the ports failed completely, but I can imagine a scenario where the ports are partially functional, but have lost their ability to autonegotiate connection speed/protocol.