Latency And File Transfer
Of course, streaming media is only one kind of data that gets used in the home. For gaming and other tasks, latency can be a significant performance factor. We measured this in a rather simple way, by running multiple ping instances at various times of day and reporting the slowest and fastest of all returned values. For example, the 5 GHz tests often usually reported back with <1 ms, but every so often we’d see a 5 ms number as conditions fluctuated. Ping won’t provide decimal results faster than 1 ms, so we represented “<1 ms” as 0.1 in our data, since it’s obviously impossible to have zero latency.
Without question, Gigabit Ethernet and Netgear’s 5 GHz kit rule the in-room test while the MoCA and powerline gear show some delay. Admittedly, 5 ms is pretty decent as far as delays go.
Statistically, the numbers in our distance test look almost identical, with 5 GHz WiFi showing no detectable latency increase, MoCA actually improving slightly, and powerline giving away only the barest of additional delays.
File transfer remains one of the biggest home LAN uses, so we amassed a 2GB collection of hundreds of miscellaneous files—everything from MP3s to Windows system files. Then we compressed this and added in even more files to reach a .zip file with a total 2GB size. A single 2GB file should take far less time to transfer than 2GB comprised of many files. But would the different connection technologies perform differently under these two transfer types? We ran this test set in our same-room configuration, pulling the files on and off of our Gigabit Netgear NAS box, in search of best-case feedback. We divided out the total transfer time to arrive at final Mb/s performance.
This is old school TCP work. With our single-file test, Gigabit blows away all comers at 230 Mb/s, but 5 GHz 802.11n surprisingly comes from behind to take a roughly 20% lead over MoCA. Powerline limps along at a mere 21.5 Mb/s, clearly suffering under this load type. With that said, powerline and Gigabit Ethernet both run here at about 1/3 of the raw TCP performance we saw in Zap. WiFi and MoCA only fall by about half—a curious difference.
Again, 5 GHz WiFi emerges as a successful underdog here, taking only a very small hit to throughput while Gigabit Ethernet got whacked in nearly half. MoCA also takes a significant step down, and powerline, already limping, drags along even more slowly. For more on why the transfer rates are dipping so low on Gigabit, check out "Dude, Where's My Bandwidth?"