EnMesh supports SMB (Samba) file sharing, but it's not the easiest thing to use. First, you'll need either slim/small flash sticks or a cabled USB drive. I had to jam the typical flash drive in the lower photo into the connector. The connector panel notch slopes upward toward the back edge, so the flash drive cover exerted some torque on the USB connector.
Connecting USB drives to EnMesh
You can plug a drive into each node and it will appear as a separate share. You must enable each drive, which can be done only in the Device settings page in the app. The web GUI Storage page only displays status.
EnMesh Web GUI Storage page
Since the port is USB 2.0, I would expect throughput in the mid 20 MB/s range at best. Since most Wi-Fi Systems don't support storage sharing, it's not one of our benchmarks so was not measured.
I decided to go back and run a quick storage test using our standard drive and robocopy script on a FAT32 formatted volume. I measured about 6 MB/s write and 7.5 MB/s read. I noticed Windows Explorer took a long time to update the folder window when drag-and-dropping a file > 1 GB.
All performance tests were done with v1.5.31 firmware and v1.32.1709261 Android app using select tests from the Version 10 Router process.
The iperf3 WAN to LAN and LAN to WAN throughput tests came in at 933 and 927 Mbps, respectively. This puts it in the top group of Wi-Fi Systems for this relatively unstressful test. The test was run with default settings, which had Firewall and all DoS protections disabled.
Routing throughput - iperf3 method
The HTTP test method proved its worth, causing EnMesh to reboot every time the 10KB filesize test was run. I experimented with disabling the DoS protections, which had no effect. But disabling the Firewall Module Function (Web GUI Firewall > Firewall Setting menu) did the trick, allowing the full test to run.
The plot below compares EnMesh with TP-Link's Deco M5 and Google Wifi, two of the more aggressively priced three node Wi-Fi Systems that also do not have a second 5 GHz radio. EnMesh performed better than the TP-Link, but not as well as Google Wifi.
Routing throughput - HTTP Score comparison - WAN to LAN
Plot key file size: [A] 2 KB, [B] 10 KB, [C] 108 KB and [D] 759 KB file
Results are very similar for uplink.
Routing throughput - HTTP Score comparison - LAN to WAN
EnMesh was run through our new Wi-Fi System test process, letting EnMesh use whichever channels and bandwidth it chose. We've been seeing many Wi-Fi Systems use 40 MHz bandwidth in 2.4 GHz, which gives them an advantage in our process, which uses a clean RF environment with no neighboring networks.
In the real world, APs and routers are supposed to fall back to 20 MHz bandwidth if a neighboring network is operating on an interfering channel. I experimented with setting up a beaconing AP to try to force fallback to 20 MHz, but found the method was not reliable. I asked eero and TP-Link to explain why their products did not appear to fall back to 20 MHz bandwidth. Only eero replied, saying they adjust transmit bandwidth on a frame-by-frame basis. I'm not clear whether this is a loophole in the 802.11 standard, but that's eero's explanation.
At any rate, while EnGenius chose auto channel selection as EnMesh's 2.4 GHz default, it also set default bandwidth to 20 MHz. Since this is the default, I didn't change it or either the 2.4 or 5 GHz channels, as per our test method. This gives EnMesh the same disadvantage Google Wifi has in our Wi-Fi System Charts and Ranker, since Google also has chosen 20 MHz bandwidth for its 2.4 GHz radio
EnMesh in test chamber
I should note a departure from the "set-to-defaults" test configuration. The first part of the test process discovers each radio's BSSID (MAC address) and operating channel, but the test client was unable to associate on the 2.4 GHz band with the default Prefer 5 GHz and -65 dBm 5 GHz RSSI settings. I had to use the web admin interface and set the Band Steering and Fast Roaming controls to Disabled so that 2.4 GHz association could occur.
Settings for Wi-Fi test
I confirmed both 2.4 and 5 GHz RSSIs were well above the -65 dBm threshold. So this feature may require more work.