The second part of wireless testing used the open air mesh test process I've been using with all mesh products. For these tests, I've chosen to compare only the three "three-pack" mesh products, i.e. eero, Luma and GWifi. As an added bonus, however, I've included test results for eero with its Version 2.0 firmware.
The first chart shows downlink performance with the middle node in the hallway location. This spot is a tougher test in that it has higher attenuation than the living room location. As a reminder, the Kitch-Reconnect test location represents two-hop results, with connection confirmed by BSSID. I should also note that the GWifi app flagged the Hallway location as having poor mesh performance and suggested moving the node.
GWifi does better than Luma in all except the best-case Office location. Comparison to eero is mixed, with GWifi doing about the same or better depending on location and eero firmware version. Luma and GWifi connected on 5 GHz for all tests shown, which seems to be the preferred band for the NETGEAR A6200 USB client I use. eero also used 5 GHz with the old firmware, but roamed to 2.4 GHz in the Living Room and Kitchen locations with the middle node in both the Hallway and Kitchen locations. That may account for the lower eerov2 throughput in those locations.
Mesh throughput summary w/ Hallway node - downlink
Hallway location uplink shows GWifi tracking lower than eero in most cases.
Mesh throughput summary w/ Hallway node - uplink
Moving the center mesh node to the Living Room location, which the GWifi app had no objection to, moves GWifi above eerov2 in most cases. But eerov2 connected on 2.4 GHz for the Living Room and Kitchen tests, moving to 5 GHz only with the forced re-association in the Kitchen-Reconnect test. But GWifi still had higher throughput for the last test.
Mesh throughput summary w/ Living Room node - downlink
Uplink results again show eero with generally higher throughput than GWifi.
Mesh throughput summary w/ Living Room node - uplink
Finally, I also ran a quick tests with a TP-Link OnHub configured as the root node, to see if an AC1900 class 3x3 radio would make a significant difference. It didn't. As you might expect, performance was limited by the GWifi's AC1200 class 2x2 radio. To get a performance boost, you'll need to build your mesh entirely with OnHubs.
Some folks will want to connect Ethernet devices to mesh nodes, using them as wireless bridges in addition to their Wi-Fi duties. All three products automatically support this, simply by plugging an Ethernet device into either port.
For the test, each product is set up with the root node in my office on the lower level of one end of my home and another node in the hallway and living room locations, one floor up. (See the first Mesh Mashup review for location details.) As with the previous tests, I just turned products on and let them link up. I used the same Lenovo x220i laptop for testing, but connected it via Ethernet to one of each product's Ethernet ports.
The downlink chart below shows performance by location. The Living Room spot has higher throughput than the Hallway location for all products, with the GWifi having the biggest jump, from 88 to 210 Mbps!
Wireless bridge performance - downlink
Uplink results again show improved throughput in the Living Room location. Luma wins the prize here for most improvement, increasing to 92 Mbps from 14.
Wireless bridge performance - uplink
These results are important because a mesh node can, at best, pass on only the throughput it receives via its backhaul link. This is one of the reasons Orbi, witih its dedicated 4x4 AC backhaul link, beats the pants off all the shared 2x2 based solutions. For reference, Orbi's results were 335 and 528 Mbps for downlink and 284 and 413 Mbps uplink for Hallway and Living Room locations, respectively.
Given its brand, Google could afford to take its time to enter the Wi-Fi mesh market. But name gets you only so far, you also need advantage in price, performance or features to move market share your way. Google definitely has the edge on price for now, although Luma has reduced its three-pack, but not its single point, price to match Google's. Given the similarity of its design to Google Wifi's, Luma is in a much better position than eero to do this and still stand a chance of making money. But given Google's deep pockets and probable volume advantage, it's questionable whether price matching is a viable long term strategy for Luma.
eero appears to have the edge on Google Wifi for performance, either with or without eero's new V2.0 firmware. Backhaul performance, as measured by the wireless bridge tests, did improve, but didn't consistently provide the 2X performance boost eero claimed in its briefing call.
For routing features, Luma, eero and GWifi are about equal, with GWifi's main disadvantage being is inability to work as access points in a mesh configuration, which is sorta the reason you buy it, eh? On the plus side, Google Wifi is definitely the one you'll want if you like to measure performance of something other than your internet connection. And its bandwidth monitoring reports are a nice plus, too.
The bottom line depends on your situation.
- Already have an eero, or have bought and returned it and are still desperately seeking mesh Wi-Fi - There's no good reason to try Google Wifi. Get a NETGEAR Orbi.
- Have Luma, but aren't happy with its performance and want to minimize your additional cost - Give Google Wi-Fi a try.
- Are looking to see what this whole mesh Wi-Fi thing is about, but are on a budget - Give Google Wi-Fi a try.