After over 30 hours testing five mesh networking kits in a large, complicated, multilevel home, we’re confident the Netgear Orbi kit is the best choice for most people—if you need a mesh kit at all. The Orbi system and other mesh networking kits are great for large and troublesome homes where a single powerful router won’t cut it, but our testing showed that most people will still be fine with our current router pick. If you do need more than one access point in your home, Orbi is the hands-down winner.
Unlike the other mesh kits we tested, the Orbi kit includes only two units rather than three, but it still had the fastest throughput and the best range of anything we tested. All the other kits we tested use the same Wi-Fi signals to talk to one another that they use to communicate with your devices. The Orbi base and satellite are connected by a dedicated high-bandwidth Wi-Fi connection that doesn’t handle anything else. The simple “Put the router where the Internet is, put the satellite in the middle of your house” instructions are impossible to screw up, and the two-unit setup also means that your laptops or phones spend less time shifting from one access point to another—or, worse, not reconnecting when they should.
Every other kit we tested relies on a smartphone app and cloud services for configuration, which means if your Internet goes down, your home network—partially or even entirely—goes down with it. Instead, Orbi uses the same well-established and functional Web-configuration interface as Netgear’s high-end Nighthawk routers. It’s the only mesh kit we tested that is fully configurable and usable without an Internet connection, and it’s also the only one that provides the full, deep feature set that technical users expect from a high-end router, including plenty of Ethernet ports on both units.
Eero, the most publicized of the home mesh kits, is a good choice for especially large homes, because its satellites can talk to one another instead of having to talk to the base and you can easily add units as you need them. In our tests the Eero kit didn’t have the raw signal strength of the Orbi or AmpliFi HD systems, but its three medium-powered units were enough to blanket our test environment in usable Wi-Fi. As with all of the mesh offerings we tried, the placement of Eero access points is critical: Doing it properly requires some forethought about where your problem areas are and how best to cover them without putting any one Eero unit too far away from the one connected to your Internet. Eero also supports an “Internet pause” feature, accessible through the smartphone app, that makes it very easy to get the attention of unruly kids, spouses, or roommates. The Eero system is extremely well-documented, too, in a style that won’t scare off network novices yet still gives more technically minded people all the details they want.
If you have a single-story house that’s smaller than, say, 2,000 square feet, a mesh networking kit might be overkill. For about a quarter of the money of most of the mesh offerings we tried, our router pick, the TP-Link Archer C7, will bathe the vast majority of houses or apartments in plenty of signal. In our testing, the Archer C7 performed as well as or better than even the best mesh kits in every location aside from our big problem areas in the half-basement, on the wrong side of a foundation slab.
Even if you don’t get the best signal in some parts of your house with an Archer C7, it might be worthwhile to just move your Internet connection to a more central location and put the Archer C7 there. In our test environment, an Archer C7 placed in the center of the house (where the Orbi kit’s single satellite went) would have worked extremely well.
I’ve been professionally testing and deploying wired and wireless networking gear in homes and businesses for the better part of 20 years. I’ve also written feature articles on networking and storage for Ars Technica and Opensource.com.
For this guide, I supplemented my own observations with reviews from SmallNetBuilder and CNET, as well as calls directly to engineers or founders at Netgear, Luma, AmpliFi, and Eero. I also checked Amazon reviews, Reddit threads, and the opinions of a few other network professionals for some details. Finally, I spent two weeks thoroughly testing all of the devices for throughput, latency, features, and general user experience in a challenging physical environment that cries out for multiple-device networking to solve its issues.
In my experience, complaints about Wi-Fi come in two basic flavors: “Why does using the Internet stink in this room?” and “Why is it so slow to copy these files?” Aside from obsessively hitting speed-test websites to see if your results match what your ISP sold you, I don’t actually see too many people complaining about the Wi-Fi being too slow for what they’re doing on the Internet. What makes people (including me) mad is when things simply don’t work—a problem that has more to do with coverage and latency (how long it takes for your inputs to make it to the other end of the connection) than with raw speed. A mesh networking kit will almost certainly solve both these problems if you’re experiencing them in your house, but it probably won’t be the most affordable choice.
You should consider a mesh networking kit if you have a large house that a single powerful router can’t cover, if you can’t move your router to a central location, and if you can’t run Ethernet cable to an inexpensive wireless access point. Even then, you should consider adding a wireless extender to your existing router. All of these options are much less expensive than tossing everything out and getting a mesh networking kit.
To be clear, a good mesh networking kit will almost certainly solve your Wi-Fi problems and give you great signal throughout your house, and it’s a reasonable option if you’re starting from scratch in a large space, or if you just want something that’s easy to set up and manage. But the cheapest good mesh system we’ve found costs $400, and most people just don’t need to spend that much.
Typically a mesh kit won’t make your network faster. Mesh can offer better coverage in a wider area, which will make your connection feel faster because your devices aren’t grabbing at faint wisps of signal—but none of the mesh devices we tested outperform the TP-Link Archer C7, our recommendation in our stand-alone Wi-Fi router guide. A network with multiple access points, like a mesh, can sometimes handle a large number of devices (smartphones, tablets, laptops, and so forth) better than a single Wi-Fi router can—but very few people’s homes or even small businesses will have more devices than a single good Wi-Fi router can handle.
Mesh doesn’t really start to shine until you don’t have wires, don’t want wires, and have lots of trouble spots (or one really big trouble spot) with poor or no coverage. Houses with old brick or lath-and-plaster walls are good candidates, as those walls kill signal and are tough to route cable through. In that situation, a mesh kit can be the right choice.
Aside from throughput and coverage, mesh kits can offer family-friendly features—such as Luma’s network filtering or Eero’s “family pause”—and ease of setup. This last point is what really sets mesh apart from other Wi-Fi approaches. Adding an extender to your existing network might involve familiarizing yourself with the UI of a different brand than your current router, configuring it to work the way you want it to, and then praying that it does. Meanwhile, a mesh networking kit promises to simply work, and the company behind it promises to support the whole thing if you have trouble.
For our first round of testing mesh networking kits, our sole criterion was simple: Does it exist? In the future we may need to rule out mesh networking kits by speed, specs, or price, but for this go-round we tested all of the consumer mesh networking kits that were available at the time (not announced, not in crowdfunding, not preproduction). That group included the Eero, Luma, and Netgear Orbi offerings, as well as both the HD and standard versions of the AmpliFi kit. Google Wi-Fi, Amped Ally, and Plume will have to wait for next time.
Before we get deep into the discussion, a quick note about rated speeds. We’ll tell you each device’s claimed speed class—such as AC3200 or AC1750—but we won’t get too excited about it, and neither should you. These ratings refer to theoretical maximum ceilings defined in the abstract specifications of wireless protocols, and they have less to do with real-world performance than the big number on a compact car’s speedometer does. Honestly, it’s worse than that: Your Civic will at least break 100 mph, but your “AC3200” router won’t ever get close to a single gigabit per second, much less 3.2 Gbps.
For the objective tests—coverage and performance—we set up each kit’s units in a challenging home environment. The two-story, 3,500-square-foot house we used is built into a hillside, and while its top floor opens onto the front yard, its bottom floor opens onto the backyard. What makes this such a tough house to cover adequately is the location of its network closet (where the Internet connection comes in), plus the foundation slab underneath half the top floor.
It gets really confusing trying to describe which device is which when you’re talking about three-device kits like the AmpliFi, Eero, and Luma systems. At my lead editor’s suggestion, I nicknamed each kit’s devices “Larry,” “Curly,” and “Moe.” In each of those kits, “Larry” is the base device that stays in the network closet, “Curly” is the first satellite residing in the farther upstairs bedroom, and “Moe” is the second satellite sitting in the kitchen, just above the downstairs bedroom.
For most of the bottom floor, a straight line to “Larry” in the networking closet goes through the foundation slab—and in some cases, through several feet of packed earth underneath it—effectively killing any direct Wi-Fi signal dead. We tested latency and throughput in each of the four corners in both floors, as well as inside a parked car in the driveway outside.
The upstairs office location is as close to a slam dunk as you’re likely to find, with a clean line of sight to the Internet connection only about 20 feet away. The closer upstairs bedroom is nearly as good, at slightly less distance but with interior walls interposed. The farther upstairs bedroom has plenty of distance and interior walls and furniture in the way but still shouldn’t be considered more than moderate difficulty in terms of coverage.
The real challenges are the downstairs floor—which is mostly blocked by the foundation slab underneath the network closet on the floor above—and the car, which is well outside the house, on the other side of a carport with a minivan in it (and which we tested with the engine running and with a smartphone streaming music over Bluetooth to the stereo).
During testing, we disabled the house’s existing wireless network, a professional-grade Ubiquiti setup featuring multiple managed access points each with their own direct-wired connection. However, we left several rogue signals in the house—including TVs, Roku boxes, and printers broadcasting their own ad-hoc networks—to do their usual useless and noisy thing, just as they probably do in your home. The neighbors also kept their Wi-Fi networks going, which left somewhere in the vicinity of 30 to 50 SSIDs visible at any given time on the top floor.
Most network-performance tests are conducted with a tool called iPerf and recorded in megabits per second, as you see on speed-test websites and the bill from your ISP. I chose instead to measure using real network file sharing and to use megabytes per second—the numbers you’ll see when copying a file from one place to another.
I chose to buck the trend because I believe in measuring where the pain is. In my experience, most people complaining about the speed of their network are disappointed at how long it takes to copy files from a server or NAS (measured in MB/s—multiply that by 8 for the equivalent in Mbps). Most people complaining about the quality of their network are complaining about latency or dropped signals, when a click or a tap takes a long time to register or nothing happens at all.
The first rule of testing: Make sure the test equipment can outperform the thing you’re actually testing. In this case I shared a folder from an SSD-backed ZFS file system with inline LZ4 compression enabled, and I dumped an unholy bunch of zeroes into 100 separate 1 MB files.
Reading those files across the Wi-Fi networks being tested does a good job of simulating a big, fairly easy file-copy run. If you copy files smaller than that, you’ll start seeing bottlenecks due to the network overhead of all the individual copy requests. Keep in mind that you can also very easily have storage bottlenecks, where the disk on one side or the other can’t keep up with the network.
Running the local-only tests first ensured that in our real tests the bottleneck we were measuring really would be the network, not the server itself. As you can see in the graph above, our tests completed locally at over 1 GB per second—more than fast enough for us to eliminate storage speeds as an issue when testing over the network.
On the client side, I used an Acer C720 Chromebook with a dual-band, dual-stream TP-Link Archer T4U 802.11ac USB 3 Wi-Fi interface. If your devices are 802.11n rather than 802.11ac, you should probably not expect to see speeds as high as the ones we show here. Before I bought the Archer T4U for these tests, the fastest Wi-Fi interface I owned was the Intel Centrino 802.11n “Ultimate” built in to my work laptop, which could pull an absolute top speed of only about 8 or 9 MB/s from any of the equipment tested.
We tested each mesh kit a minimum of three times at each location for both latency and throughput, and then we selected median results by throwing out the top and bottom runs. This step was necessary because wireless networking is messy: A wired Ethernet connection is like an intimate date at a quiet, candlelit table. Wi-Fi is more like two blue whales bellowing across the ocean. As a result, something as simple as the cat running across the room can visibly affect the numbers when you’re obsessively watching a throughput test.
Finally, we tested plain-and-simple usability by browsing random selections from the front page of YouTube on a smartphone in each of the test locations for a few minutes. It’s hard to entirely capture a subjective experience with numeric tests, but when your signal drops for five seconds at a time and the interface stops working or your video starts buffering, you know it—so this test was our fallback to make certain whether the kits we’re recommending will or won’t work at the iffier extremes of coverage.
Our picks are below. If you’d like to skip down and read the test results first, click to jump to the Results and Analysis section.
Netgear’s Orbi is the best Wi-Fi mesh system for most people because it covers all the bases in the simplest way. It was at or near the top of the pack of mesh devices in every location we tested, with far less time and effort spent on our part playing around with it or trying to get it to do its best. It’s easy to set up and use, it has all the advanced features you might expect in a high-end router plus plenty of Ethernet ports, and unlike most mesh kits, it still works without an Internet connection. It’s also the only kit we tested with a dedicated backhaul band, one that is reserved only for communication between the router and the satellite unit.
The Orbi base and satellite devices both look like small, somewhat oddly shaped white plastic vases, with a “halo glow” at their top that turns on only during setup and reboot. Each is classified as a tri-band AC3000 device, with 1733, 866, and 450 Mbps streams. The 1733 Mbps number refers to the dedicated 5 GHz backhaul channel that the Orbi satellite uses exclusively for communicating with its base router; the 866 Mbps and 400 Mbps figures refer to the two-stream 5 GHz and 2.4 GHz bands the Orbi system uses to communicate with your devices. You should not mistake these numbers for real-world speeds, but they do add up to a very respectable chunk of Wi-Fi bandwidth that’s carefully allocated to be just as usable on the satellite as it is on the base.
Ease of setup is a pretty big deal with Wi-Fi mesh, because the placement of the units can be as important as which kit you buy. It might be easy to look at the two Orbi devices versus the three-device kits sold by other brands and think “Well, I’m getting less for my money,” but that’s not the case here. The fact that the Orbi kit has just two units instead of three, with dead-simple instructions to put the satellite in the middle of the house, will ease the technical anxiety of people who simply want to plug it in and get on with their lives. And as you can see in our testing results, the Orbi kit actually does more with its two units than other systems can muster with three. If you really need more hardware, you can purchase and add another unit. This add-on connects to the base unit; Orbi isn’t currently capable of multi-hop topology. We need to stress, however, that very few homes will need a third Orbi unit.
In addition to being easy to set up, Orbi keeps technophiles happy as well, by offering four wired Gigabit Ethernet ports on both devices.1 (Eero units have only two ports.) Orbi also has a familiar and complete Web UI containing all the technical features that Netgear’s high-end Nighthawk routers have been offering for years, including static routing, advanced port mapping, firewall rules, and backup. It even has MU-MIMO support for the few devices designed for it.
The Orbi kit is the only one we tested that we can thoroughly recommend to people with patchy, intermittent, or no Internet service. The other kits lean heavily on cloud services and smartphone apps for installation and configuration, and you can’t set them up without an already-working Internet connection. In contrast, you can (and should) set up your Orbi system with a local Web interface, so it works just fine with any Internet connection or none at all. Orbi can use the same Netgear Genie app as other Netgear routers, but you don’t need that app, and for the most part you’re better off without it.2
CNET’s Dong Ngo reviewed the Orbi system in early October and came to similar conclusions: It’s easy to set up, it performs well, and it provides good coverage. Tim Higgins at SmallNetBuilder posted a review early in November that includes a complete (and pretty cool) physical teardown.
Compared with most Wi-Fi devices, the Orbi devices are a bit large and … odd. Their “alien vase with no flowers in it” design won’t offend delicate sensibilities as badly as the typical bristling-with-antennas router monstrosity, but it’s still not exactly beautiful. You also don’t want to put an Orbi unit anywhere somebody might spill their coffee on it, since the top of each device is largely an open vent—a drink spilled on the Orbi is a drink spilled in the Orbi.
The Orbi base unit and satellite both have USB 2.0 ports, but those ports don’t do anything yet. Most routers with USB ports let you use them to connect hard drives or printers, but until the USB ports are enabled, Orbi can’t do that.
Finally, if you’re really into having a smartphone app for your router, the Orbi system might disappoint you. Netgear Genie is available, and it works, but it’s unlovely and unnecessary. (For me, this is a feature, not a bug. I like my control interfaces on a real computer, where I can see them and interact with them more efficiently.)
This is all really nitpicking. I like the Orbi kit. It worked right out of the box, it kicked butt, and it didn’t ask me to give up any of the features I’m used to from techie-oriented routers.
Eero is the first and most famous of the mesh networking kits, and it’s our second-favorite of the five we tested. For most people, it’s hard to justify the $100 bump over Netgear’s Orbi, which in our tests tended to outperform Eero in almost any situation. But if you don’t mind the cost and find its unobtrusive looks, clean smartphone app, innovative features, and expandability exciting, the Eero system can still be a solid choice for your large or hard–to–Wi-Fi home.
In our testing environment, Eero was reliable and easy to set up and use. The units feature an attractive, sleek design that won’t look out of place in any decor, and the system offers good throughput and range plus serious expandability. It also provides a family-friendly “Internet pause” feature that generated a lot of buzz in its initial campaign, and its founder Nick Weaver says the company is still evolving the list of things Eero is and does in newer firmware updates—so more may be yet to come.
Each Eero unit is a dual-band, dual-stream AC1200 router that features two Gigabit Ethernet ports you can use to connect client devices or to connect the Eero units together. (You’ll also find a USB port, which works only for diagnostics.) This design means that if you already have network wiring in your place, you can use Eero in whole or in part as a simple set of access points rather than a fully wireless mesh—though if your whole house is wired for Ethernet already, you may be better off with cheaper access points.
You can buy more Eero units and adopt them into your existing network, if your place is big enough and your pockets are deep enough to justify that. Additional Eeros are available individually or in two-packs at a higher cost per unit, or you can buy another three-pack just like your first one. (We’re not sure if the system has a theoretical maximum, but Nick Weaver, Eero’s founding technologist, told me that he has seven of them in his place.) This expandability makes Eero the best choice for very large houses that aren’t prewired for Internet, or perhaps those homes with lath-and-plaster walls that block most Wi-Fi signals.
For extremely long buildings with the Internet connection on one end, you can set up your Eero system in a “multi-hop” configuration where the third unit repeats the signal from the second unit, instead of the base unit. When I tested the multi-hop configuration, latency was good, at less than 15 ms average and less than 40 ms peak, and throughput at 20 to 25 Mbps is plenty for streaming video over the Internet—but the raw speed won’t impress for any local, high-speed tasks like sharing files larger than a few megabytes at a time. Basically, the connection will work reliably at acceptable speeds, but not very quickly.
A note: Eero devices are physically identical, but after you set them up, it does matter which one is which—particularly which one you decide to connect directly to your ISP’s modem or router. You name your Eero units in the smartphone app during setup, but physically labeling them to match is highly recommended. If you don’t do so, you’ll be in for a hassle if they ever get unplugged and moved around.
SmallNetBuilder’s Tim Higgins provides an exhaustive review of the Eero system’s physical innards and technical performance, and CNET’s Dong Ngo covers some of its pros and cons, ultimately preferring standard Wi-Fi gear.
Note: After this guide went to production, Eero announced a firmware update that the company claims can improve performance by up to 40 percent. We’ll retest the Eero soon and update our results.
The closer upstairs bedroom and the upstairs office were both “slam dunk” sites from the network closet. The office has a clean line of sight, with no walls or doors, from about 25 feet away, and the near bedroom has a straight 20-foot shot with two perpendicular internal walls. Easy peasy. This is as good as it gets. When you’re this close to the Internet connection, with such a clean angle, you don’t need satellites—your device can (and should) connect directly to the base unit itself.
The only kit to show a significant drop in throughput between these two sites was the Luma kit, which performed significantly worse in the upstairs office. It’s hard to say whether this result was due to the few extra feet of linear distance or just Luma being inconsistent. I frequently had difficulty getting the Luma system to run properly at all, and I spent more time coaxing it to cooperate than I did while testing any two or three of the other kits combined.
The farther upstairs bedroom is both a test location and a satellite-placement site. It’s about a 40-foot shot from the network closet, with a few interior walls and cabinets at odd angles to line of sight along the way. For the three-unit kits—Eero, Luma, and the two AmpliFi sets—I placed the first satellite (nicknamed “Curly”) on a writing desk against the back wall, to provide a good clean angle around the foundation slab to the downstairs test sites. (I didn’t place Orbi’s single satellite here; I followed Netgear’s installation instructions and put it as close to the center of the house as possible.)
The signal here should have been quite good for any reasonably powerful device in the network closet—but as you can see in the graph above, the Luma and Amplifi AFI (base model) kits didn’t look particularly powerful.
For every kit, I disconnected and reconnected my laptop to the Wi-Fi to give it every possible chance to connect to the nearby satellite. This tactic worked immediately for everything but the Luma system: My laptop stubbornly refused to have anything to do with the Luma satellite sitting on the same table. This problem was especially perplexing because it showed as connected in the Luma app, the satellite’s MAC (hardware) address showed up attached to the SSID with extremely high signal on my Wi-Fi Analyzer app, and I had no trouble connecting to this specific Luma satellite from other locations, where it wasn’t as desirable.
In the graph above, the numbers illustrate that, for most of the kits, it would have been better to put the satellite a bit closer to the router. The Luma kit in particular struggled badly here, even when I tried placing its first satellite 12 feet and one wall closer to the network closet, in the same location as the Orbi system’s satellite. Luma just did not provide a powerful enough signal to handle this house.
The problem is, there’s no way to figure out this kind of placement issue without obsessively testing, adjusting, and retesting. I carefully measured signal strength (in dBm, or decibel-milliwatts) before choosing sites, and I still ended up discovering that I could’ve done better. I ended up taking a few hours of testing for each kit, as well as some more time assembling those results into graphs and analyzing them, before I figured it out. Throwing out those results and testing everything again might have made some of the kits look better on paper, but it wouldn’t have improved the real-world experience of using them as they’re intended to be used.
I also avoided placing satellites in locations that had excellent signal but that I would never consider for day-to-day placement in a home I had to live in. At the end of the day, our tests had to be a meaningful reflection of real people’s experiences, and that meant living with the initial good-but-not-perfect satellite locations. Our experience highlights the Orbi system’s biggest real-world appeal, thanks to its two high-powered units and simple instructions that work extremely well—you just put the satellite in the middle of the house, and let the rest take care of itself.
Outside the house
The long-distance shot to the car was a real torture test. We had the car about 55 feet from the network closet, with two interior walls, a brick exterior wall, some furniture, and a parked minivan in between. To make the test even tougher, I ran the test sitting inside the car with the engine running and with music streaming from my phone to the car stereo on Bluetooth.
Both AmpliFi kits, the Eero system, and the Luma kit offered the option of connecting either to the base unit in the wiring closet or to the unit we called “Moe,” the satellite sitting in the kitchen—a mere 40 feet, one interior wall, and one exterior wall away. Our devices seemed equally likely to pick either satellite or base, with no real difference in performance. Luma was only intermittently visible by its SSID, and it was completely unable to connect for actual testing.
The Orbi system didn’t move data as quickly as the TP-Link Archer C7 router did here, but the connection was notably solid and usable. Both Orbi and AmpliFi HD seemed determined to connect on the 5 GHz band, even though 2.4 GHz was a better fit for such a remote location because of its greater range. That behavior allowed the AmpliFi AFI and Eero systems to outperform their higher-powered competitors on this test—those two couldn’t even pretend to punch a 5 GHz signal from that far, so they connected on the more appropriate 2.4 GHz band every time.
The TP-Link Archer C7 router really demonstrated its range and power here: On the 5 GHz band it pushed nearly 9 MB/s to the car. Throughput isn’t everything, though, and even the Archer C7 is really better off at 2.4 GHz at such distances—although it made bigger numbers on its 5 GHz band, I spent a good five or six seconds staring at no throughput at all before it lurched into gear. But at 2.4 GHz, the Archer C7 produced immediate, solid throughput and sustained that throughout the test.
To be clear, Orbi does offer the ability to separate its SSIDs, just as I did on the Archer C7. I expect that the Orbi system would have performed much better in the car if I had forced it down to the 2.4 GHz band. The fact that I didn’t need to mess around with Orbi’s fine details is a feature, not a bug. It was still usable for anything I would want to do while sitting in a car in the driveway—browse YouTube or Facebook, download music from Google Play to my phone’s internal storage, and so forth. Not having to mess around with Orbi to get it to do what I needed it to, here or elsewhere, is the biggest single reason it became our top pick.
Moving downstairs, you can start to see why this house really needs mesh. Our high-powered Archer C7 router sitting in the network closet had about a 40-foot shot through a couple of interior walls, some furniture, and a floor/ceiling at a nasty oblique angle—though it did at least manage to miss the foundation slab here. As a result, the Archer C7’s 5 GHz band was completely obliterated, but on its 2.4 GHz band it managed to deliver a solid, surprisingly reliable 2.4 MB/s.
This test was a much tougher shot for the Orbi system than for the three-device kits, since it needed to clear 25 feet, some furniture, a still-pretty-oblique floor/ceiling angle, an interior wall, and some closet space to get there from its satellite atop the island in the living room. It managed the task gracefully, though, with a solid and reliable 4.5 MB/s.
The Eero kit—and, surprisingly, the Luma kit—ended up shining here. We placed their respective “Moe” units in the kitchen upstairs, almost directly above the downstairs bedroom’s test station. With less than 15 feet of distance and nothing but one cabinet and a pretty clean perpendicular shot through the floor/ceiling in the way, they were able to outperform the higher-powered but more distant Orbi.
If you’re wondering why the AmpliFi and AmpliFi HD kits didn’t do better even though they also had a satellite in the kitchen, I’m going to blame it on their curious physical design. AmpliFi’s satellites have particularly limited placement options, since they plug directly into power outlets. Worse for my purposes, the designers clearly expected those outlets to have the ground prong on the bottom. In our test house, the ground prong is on the top of the outlet, which means AmpliFi’s satellites are not only stuck at ground level but also very nearly dragging on the ground—or in this case, stuffed directly behind a microwave on a counter, tilted at a weird angle from their base in order to fit there at all.
It’s impossible to get a signal to the downstairs office from the network closet without going through a concrete foundation slab. I was frankly astonished to see the Archer C7 manage to get a connection at 2.4 GHz anyway—any other router I’ve tested has been lucky to get enough signal through for the SSID to show in a list, much less connect and move data. Still, don’t get too excited—that 400 KB/s or so was not usable; it included several periods of 10 or more seconds of completely dead air and reconnection. I wouldn’t even have bothered to let it finish the test, if I hadn’t been so impressed that it could finish the test at all, however unusably.
Luma did oddly well on this test. When my laptop was in the upstairs bedroom right next to Luma’s “Curly,” I couldn’t get my laptop to connect to Luma for the life of me; it invariably chose to connect to “Larry” (the base router) or to “Moe” (in the kitchen) instead. Down here in the office, though, it connected—eventually—to “Curly,” and the connection did a reasonably good job afterward. What doesn’t show in the numbers is that it required several attempts to connect to the network in the first place.
Once again, Eero outperformed AmpliFi HD despite its lower-powered units, simply because they didn’t need to be crammed into an outlet at floor level. The base-model AmpliFi failed to excite—and wasn’t especially reliable in actual use—and the Orbi system stole the show with its combination of extreme high power and good placement options.
The downstairs bath was another torture test. Hugging the edge of the top floor’s foundation slab, any shot from the network closet to this room required punching through the slab plus a few feet of earth. Once again, the TP-Link Archer C7 router shocked me by managing to get a 2.4 GHz signal through and—however painfully—eventually completing its test run.
The Eero, AmpliFi HD, and Orbi kits all made it look as if we were really supposed to have Wi-Fi in here. Unfortunately, the Luma system failed miserably, and the base-model AmpliFi posted relatively decent numbers at first glance but did not actually satisfy in terms of reliability and usability.
Multi-hop (satellite-to-satellite connection)
So far we’ve looked only at the placement of satellites with direct Wi-Fi connections to the base router. Eero and Luma also have the ability to connect in a “multi-hop” configuration, with one satellite connecting to another satellite and then to the base device, an arrangement that’s useful in long buildings with the Internet connection at one end, or with solid walls that block Wi-Fi signals. I would have liked to test Luma in a multi-hop configuration, but I spent so much time getting it to work reliably at all that I didn’t have time to play with it in a more advanced topology.
Eero‘s “Moe” unit connected effortlessly when I moved it from the kitchen upstairs to the downstairs office, almost directly beneath its “Curly” unit. And the experience was very smooth and reliable. However, client devices connected only at 2.4 GHz—which I assume was by design, to allow the second satellite to communicate with the first at 5 GHz without having to worry about a client “talking over it.”
This was a very reliable and solid connection—latency was good at less than 15 ms average and less than 40 ms peak, and throughput at 20 to 25 Mbps is plenty for streaming video over the Internet—but the raw speed wouldn’t impress for any local, high-speed tasks like file sharing. If you need to move more than a few megabytes at a time, you’d be better off dragging the files to a portable hard drive and then walking them to the other computer.
I wouldn’t recommend placing Eero devices where they need multi-hop communication unless you honestly need to. The Eero system works much better when all satellites can connect to the base unit. If you do need to bounce signals like in a pinball machine, though, Eero’s got you covered.
Wired versus Wi-Fi
As exciting as all this new high-powered Wi-Fi tech is, you shouldn’t ever mistake a Wi-Fi network as offering anything close to the performance a wired connection offers. Even in the absolute best conditions for Wi-Fi, a standard wired connection is in a completely separate league.
As you can see in the graph above, the wired connection was more than five times faster than the highest throughput achieved by any wireless alternative at any location during this test. Or if you have a flair for the dramatic: It was literally faster than every other option put together.
The picture doesn’t get any prettier when you look at latency. Avid Internet gamers, take note: The absolute best Wi-Fi connection you can manage, with the absolute best Wi-Fi gear you can find, will significantly increase your ping time.
The spread between average and maximum latency is also worth a look. When you see a big difference between the blue and red bars, that means your connection is unpredictable—in other words, the dreaded “lag spike” is happening before you even leave your network. The wired connection, on the other hand, deviates by only 0.04 ms between average and worst-case latency. Your reflexes are definitely not that good.
The message here should be clear: If you really care about maximum performance, you need a wired connection, no matter how good your Wi-Fi gear is.
The AmpliFi HD system produces a more powerful signal than the Eero kit at a significantly lower cost, and unlike the Eero offering, it comes from a company with a long pedigree in enterprise mesh networking. Like Eero, AmpliFi HD has a polished smartphone app. This feature should have made it a slam dunk for our first runner-up pick. Unfortunately, AmpliFi’s awkward satellite design makes it much less compelling in reality.
The AmpliFi HD kit consists of a small, cube shaped AC1750 base unit with a digital display and two tower-shaped satellites that look like something the Star Trek: The Next Generation crew might have used for Wi-Fi if it hadn’t been built into the ship. The satellites—which contain dual-band, three-stream AC1750 Wi-Fi radios—offer no wired ports and can connect directly only to the base unit (not to one another, despite implications to the contrary). They also have no power cords: They plug directly—and awkwardly—right into the power outlet.
The magnetic breakaway coupling holding the satellites’ antennas on will prove irresistible to small children, mischievous roommates, and maybe even the occasional dog. Anyone can easily grab the antenna segment and split it from its base, and the need to plug the base directly into a power outlet means it is both always in reach and poorly placed.
Aside from the potential problems of someone wandering off with half the satellite while the other half stays plugged in, the baseboard-level placement of most power outlets means that the satellites have to punch through every conceivable obstacle on the way to your laptop, tablet, or phone. This situation has a very real impact on the quality of your Wi-Fi when you’re connected to a satellite instead of directly to the base unit. It also means that the fascinating-to-break-apart satellites remain well within reach of even very small children with inquisitive hands.
AmpliFi HD features a polished and functional smartphone app that provides simple setup and configuration, as well as the ability to “pause” Internet connections for the entire network or for an individual client device.
The base-model AmpliFi unit looks just like its bigger sibling, the AmpliFi HD base, but this offering lacks AmpliFi HD’s mitigating high signal strength. Instead of the three-stream 802.11ac satellites that come with the AmpliFi HD kit, the base model set ships with dual-stream 802.11n mesh points. AmpliFi’s specifications state that the HD kit features 26 dbM transmit power, compared with 24 dbM for the AFI version. That doesn’t sound like much of a difference—and if such specs are accurate, it really shouldn’t be—but in our testing, the AFI system’s lower signal strength had a significant negative effect.
At $200 at the time of our testing, AmpliFi AFI is the least expensive mesh system we tested, but that still leaves it more than twice as expensive as the TP-Link Archer C7 router, and therefore a dubious value. The Archer C7 was tremendously faster in all upstairs locations, and the AmpliFi AFI kit really didn’t do all that well downstairs. It’s no surprise that AmpliFi wanted us to test the HD model instead.
If it had been possible to place the AmpliFi AFI satellites in places that aren’t ankle-level power outlets, it would likely have proven to be a much better option. We’re hoping that a future model will solve this problem, and that we can reevaluate both this version and the AmpliFi HD system without their current Achilles’ heel. For now, though, it’s hard to picture an environment that would be better suited to the AmpliFi AFI setup than to the even-less-expensive Archer C7.
Like Eero, Luma is a still-evolving three-unit mesh networking kit that promises a new approach to Wi-Fi, with simpler setup, more features, and higher performance than a single router offers. Each Luma device is a dual-band, dual-stream AC1200 router with two wired Gigabit Ethernet ports, one of which you can used to connect to the Internet, and the other of which you can connect to an Ethernet device like a desktop PC or TV. (Each one also has a single USB port, which works only for, uh, charging things?) Another similarity to the Eero system: The Luma nodes are physically identical to one another (there’s no dedicated base or satellite), so you should absolutely label them physically during the setup process. If you don’t, you’ll have a heck of a time figuring out which is which (and it will matter) should they ever get unplugged and moved around.
At this writing, the Luma three-pack is the same price as the AmpliFi HD kit and considerably cheaper than a pack of Eero devices. Unfortunately, this set also more complicated and less reliable than either, and its software needs some work to live up to its advertised features.
Among the kits we tested, the Luma devices were the lowest powered (in terms of raw signal strength), which hurt the system badly in performance and even more in usability. The Luma kit did a significantly worse job of covering our large and difficult test environment than its competitors did.
Luma offers many extra features that most network kits don’t, including an Internet pause like Eero’s, fine-grained parental controls, and what Luma refers to as “network security scanning.” The Internet-pause works fine, but the filtering and scanning claims need to be taken with a large grain of salt.
Both the parental controls and the “network security scanning” are really just DNS-based filters. This means that any attack (or any naughty content) that doesn’t rely on DNS resolution will sail right through. I demonstrated this flaw very graphically during testing just by typing “hardcore porn” into Google Image Search. I’m 44 years old, and even my mother would probably prefer I not see the resulting images. Sites like Reddit, Imgur, Flickr, and Tumblr, which are not pornographic sites in general but have areas devoted to explicit pornography, were also completely unaffected.
Luma’s “network security scanning” doesn’t actually scan your network or devices at all. It’s just a DNS filter, like the parental controls. Despite the misleading name, though, it is a worthwhile added layer of security. In my testing, Luma successfully stopped a Locky malware executable from reaching its command-and-control servers. Without that connection, the Locky invader was unable to fetch a key and encrypt all my test machine’s data with it.
Unfortunately, Luma didn’t—and couldn’t—do anything to stop me from downloading the malware itself (which came through email). It also gave me no notification, even though it recognized that a machine on my network was frantically trying to reach known criminal malware networks. And even when I visited the security tab in the Luma smartphone app, the screen showed me a big, cheerful green shield and said “Network Secure”—which it clearly wasn’t.
Luma’s founding engineer shared some screenshots from an upcoming firmware update with me, and that update does begin to address some of these issues. Luma had also already fixed a documentation issue I reported during testing—so its team is definitely paying attention, and I hope that Luma will continue to improve.
If you’re a more technical sort, you might be wondering how this stuff compares against a more traditional multiple access point stack, like Ubiquiti’s UAP line of devices. Unlike the mesh kits we tested, each UAP unit connects to the network through a wired Ethernet port and provides a Wi-Fi signal in their physical area. Technically, this arrangement is a much simpler option than mesh, as you don’t have to worry about a signal to the access points, just the signal each one provides. You also don’t need to worry about broadcast chatter in between the access points, or any of a host of other potential problems; you just plug a unit in (to both the power and the wired network) wherever you want some more of that sweet, sweet signal.
Aside from the Ethernet cabling, the only real downside to Ubiquiti’s UAPs is the need to install the UniFi controller application on a connected computer; PC, Mac, and Linux are all supported, and you’ll need it only for setup. (You’ll still need a router, too, since the UAPs are just access points, but your current router will work fine for this purpose.) UniFi is actually pretty cool. It lets you adopt and place all of your UAP access points, see who’s connected to them (and boot them off individually for grins, if you feel like it), and more. It’s not really that hard to use, but it definitely feels “techie” enough to scare off some less-technical people.
Our test environment was already wired for Ethernet throughout the house, and its Wi-Fi was (and still is) being provided by a several-years-old pair of the original wireless-n Ubiquiti UAPs. One AP sat in the network closet upstairs, and the second one resided in the downstairs office.
The UAP units in the house were originally priced similarly to their current-model replacement, the $80 UniFi AP AC Lite. Unlike the dual-radio AC Lite, though, our original-model UAP units are simpler, 2.4 GHz devices—which doesn’t keep them from doing a really good job.
At first glance, you’ll notice in the graph above that the Orbi system produced the largest numbers in this comparative test. (It had better do so—Orbi units contain 5 GHz 802.11ac radios, whereas the UAP devices are stuck at 2.4 GHz 802.11n.) But despite having the biggest numbers overall, the Orbi kit actually lost to the older, cheaper UAP units in four locations while outperforming them in only three.
The UAP devices did significantly better in the downstairs locations and in the car, too. Unlike with the Orbi system, their performance was near-identical in every test location but the car. And they provide such consistent performance on a single band, which means you have fewer chances for a client to stubbornly connect to the wrong band.
As we mentioned earlier, above and beyond the Orbi kit’s top throughput, we like Orbi because its fewer devices mean fewer chances for a client device to screw up. The same concept works in the UAP system’s favor.
So far, so good—but throughput doesn’t tell the whole story. Latency still tends to be a better measure of how your Wi-Fi network feels in the seat of your pants, and in terms of latency, the two UAP units absolutely stomped the Orbi system (and every other kit we tested) in every location. The only place average latency climbed higher than 5 ms was our long-distance torture test to the car, and even then it was less than 10 ms.
You might be asking: What about maximum latency? Any weird hitches or glitches or “bald patches” every now and then where nothing seems to happen?
Nope. Aside from the numbers being bigger, you might as well be looking at the average latency graph again. In this regard we were still under 5 ms almost everywhere, and even the car test was still under 25 ms.
Subjectively, you really feel this difference in heavy use. The UAP system delivers a very smooth and reliable experience with no frustrating pauses or dropped signals. In our tests, even in the car, where throughput was only a third what it was in the house proper, it connected rapidly and transferred data smoothly and consistently. You feel like you’re supposed to be connected to it, as opposed to feeling like you merely managed to connect to it.
Our original UAP units cost about the same as the newer UAP AC Lite devices linked above, making the pair of them about $160. As mentioned earlier, you do need an actual router (wired or wireless) behind them, but we won’t count the cost of that here, since whatever you’re already using will work fine. Even if you add in labor for a professional to run an Ethernet cable from the network closet to the downstairs office, the total cost is still well under $300. (Eero is also capable of using a wired connection to any or all of its satellites, and Luma at least should be capable, but they’re considerably more expensive.)
Long story short: If you can run Ethernet cables and use standard access points instead of a mesh kit, you should consider it.
Unless you have a large house with a complicated layout or walls that kill your Wi-Fi signals, you may not need a mesh network at all—you might just need to upgrade your router to something newer and more powerful. We’re mentioning the TP-Link Archer C7 last in this guide because it isn’t a mesh network option. It’s just an inexpensive, especially high-powered Wi-Fi router, the same one we’ve been recommending for years. But last doesn’t mean least: The Archer C7 is dead simple to set up, and you don’t have to worry about laptops, phones, or tablets trying to connect to the wrong AP. And did we mention it’s inexpensive? At less than half the cost of the cheapest mesh device we tested, and less than one-fifth the cost of the most expensive, it hung with—or more frequently, outperformed—all of them in the majority of our upstairs test locations.
If you aren’t certain you need mesh, or if you don’t have the deep pockets to spend on the technology, you should probably go with the Archer C7 instead.
New mesh networking kits are launching all the time, and we’ll get to them as soon as we can. For a later update to this review, we’ll be testing the Google Wifi, Amped Ally, Plume, Linksys Velop systems, and Asus HiveSpot, and we’ll also take a look at what you can expect if you add an extra Orbi satellite to our current pick.
(Photos by Michael Hession.)
Originally published: November 30, 2016