Author: Colin Dean
I put the WD MyNet N900 Central through a series of benchmarking tests. My primary comparison was with a Buffalo WZR-HP-AG300H. Most of my testing was conducted between a mid-2012 13″ MacBook Air and a late-2009 13″ MacBook Pro.
I have a 150 Mbps / 65 Mbps Verizon FiOS connection, and I get the advertised speeds. I have maxed it out on several occasions, most recently when building a new rig and reinstalling stuff. For all tests involving Internet speed tests, I used Speedtest.net with a server in Reston, VA since my normal Baltimore server was’t able to keep up during this testing cycle. These servers are known to me to be some of the few near me that can actually max out my connection.
First, I used Speedtest.net to conduct a simple bandwidth throughput test. This is a test a new user would run, so its importance cannot be underestimated.
WD My Net N900 Central (Wired)
Buffalo WZR-HP-AG300H (Wired)
WD My Net N900 Central (2.5GHz)
Buffalo WZR-HP-AG300H (2.4GHz)
WD My Net N900 Central (5GHz)
Buffalo WZR-HP-AG300H (5GHz)
I am not impressed with the upload speeds of the WD router. I did disable QoS for these tests. The download speeds were sufficient, with the 5 GHz channel on the WD getting close to maxing my connection.
Next, I executed a barrage of throughput and latency tests using iperf and a GUI for it, jperf. I recently created the OS X app+dmg packaging for jperf! I conducted tests with my MacBook Air as the client and the MacBook Pro as the server.
|Test||WD MyNet N900 Central||Buffalo WZR-HP-AG300H||Comments|
|Wired to wired, TCP throughput||928 Mbps||932 Mbps||Negligible difference. Both are gigabit-class.|
|Wireless 5 GHz to wired, TCP throughput||133 Mbps||132 Mbps||Equivalent performance.|
|Wireless 2.4 GHz to wired, TCP throughput||61.4 Mbps||81.2 Mbps|
|Wireless 5 GHz to wired, UDP 1 Mbps average latency||0.140 ms||0.238 ms|
|Wireless 2.4 GHz to wired, UDP 1 Mbps average latency||2.036 ms||11.588 ms||The WD shot up to 4 ms in some tests, but repeated testing hovered around 2 ms|
|Wireless 5 GHz to wired, UDP 300 Mbps throughput with average latency||84.3 Mbps / 0.492 ms||166 Mbps / 9.558 ms||Here we see some testing oddity. This test blasts UDP packets at the theoretical fastest I’d expect my laptop’s hardware to be able to dish out. Beats the TCP throughput of both routers.|
|Wireless 2.4 GHz to wired, UDP 300 Mbps throughput with average latency||79.9 Mbps / 0.389 ms||85.4 Mbps / 3.925 ms||The WD handles latency better all the way around.|
|Roaming Wireless 5 GHz to wired, TCP throughput, Location 1 (same room, 5 ft away)||142 Mbps||160 Mbps|
|Roaming Wireless 5 GHz to wired, TCP throughput, Location 2 (next room, 15 ft away)||120 Mbps||101 Mbps|
|Roaming Wireless 5 GHz to wired, TCP throughput, Location 3 (living room, 15 ft vertical away)||151 Mbps||157 Mbps||There’s not really enough of a difference here to declare a winner.|
|Roaming Wireless 5 GHz to wired, TCP throughput, Location 4 (kitchen, 25 ft diagonal away)||60.5 Mbps||35.8 Mbps||This is a tough location because of all of the electrical work, walls, appliances, and other devices in the way. The WD pulls through with nearly double the throughput.|
|Roaming Wireless 5 GHz to wired, TCP throughput, Location 5 (basement den, 30 ft vertical away)||71.7 Mbps||80.8 Mbps||Similar to the kitchen in opportunities for interference and signal degradation, my basement den is where my gaming rig and systems live.|
One note: there were other devices using the Buffalo’s 2.4 GHz network at the time of the review. I tried to turn off unnecessary devices, but some are required to be online at all times. This may have impacted performance for that test.
I would have liked to see better results from a router that is significantly newer than the Buffalo router. It held its own, though.
Being that it’s a NAS device, too, this review would be incomplete without some benching of the 2 TB internal drive. Using simple rsync to transfer a ~800 MB file, I saw write performance up to 8 MB/s wireless on the 5 GHz band and 16.63 MB/s when wired.
I used blackmagic to do some full-stack throughput testing. First is wired performance, the second is wireless 5 GHz.
I should note that the router crashed at one point during this benchmark. It stopped responding to wireless and wired connection requests.