Update: This equipment worked less effectively than I had thought when I originally wrote this review. See this post for an update.
Update 2: Since some people don’t seem to be interested in clicking the above ‘update’ link, I’ll put some detail here. This hardware is crap, don’t buy it. Not only is it ridiculously slow, it drops connections all the time making your network useless for transferring large files (or streams). I unloaded the powerline gear and replaced it with Multimedia over Coax (MoCA) hardware, which allows you to use your Cable TV coax for networking (in addition to TV).
When we first moved into the new house, I initially set up a single WRT54G access point (running dd-wrt) in the office in the back of the house. I quickly learned, however, that this location didn’t provide much signal to the rooms in the front of the house, most notably, the living room. Fortunately, I had a second WRT54G lying around as well as a good place to hide it in the dining room, so I set up WDS and used it to extend the network to the front of the house. This has been our basic network configuration for almost two years now. The equipment in the front of the house which requires network, such as the Xbox 360 and the Tivo, all have wireless adapters which are served by this second access point.
This was mostly dandy for your everyday bullshit web browsing and checking email. But as I started to put more demand on the network by pulling HD content off the Tivo, renting HD movies on the 360, as well as things like Netflix streaming, it started to creak. Truth was that not only was the signal between the office and the dining room pretty lousy, but just using WDS halved the bandwidth available to wireless clients. It was time to find a better solution.
The obvious solution is to use wired ethernet to distribute the network around the house. Unfortunately, our condo is the middle floor of a 3 story house, which means no access directly below or above us to easily run Cat5. While it wouldn’t be impossible to get some Cat5 snaked through the walls into the basement, it wouldn’t be easy or fun. If I took on this project, it would turn into a complete mess. If we hired someone to do it it’d likely be prohibitively expensive. Neither Corinna nor myself are interested in running Cat5 visibly around the house, so that’s not an option either.
Unable to come up with a true solution to this problem, I shelved it, accepting the status quo. A few days ago it occurred to me that there had been some companies developing equipment to send data over the electrical wiring of a home. Reviews on this equipment seemed to be pretty mixed, though. Some people raved about how brain-dead simple it was and how well it worked. Others complained it didn’t work at all. After some hemming and hawing, I decided to just give it a shot and if the equipment sucked I could always just resell it. I ended up ordering the Netgear HDXB111 Powerline HD Plus kit. It advertised a top speed of 200Mbps, but of course, actual data rates may vary…
I received the equipment yesterday and spent part of the evening doing some basic testing. First off, the actual setup of this equipment is just as brain-dead as advertised. I plugged each of the units in on opposite sides of the house and within a few seconds had link. That was it, it just worked right out of the box, no configuration, software, or anything else required. There was one other optional, but important, step, which was to ‘randomize’ the encryption keys used by the device. Out of the box, the key is set to a default value, but there is a ‘pairing’ procedure which generates a ‘random’ key and distributes it among the devices. You hold the “Security” button on the first device, and it goes through some motions and decides it is the primary device and generates a key. Then you hold the button on that device again, a light starts blinking, and you have 30 seconds to hold the button on another device. This distributes the key to the second device and you are, in theory, more secure. It apparently uses 3DES, but that requires trusting Netgear, so as always, it’s in your best interests to never rely on it. There is also some Windows-only (ick) management software you can install if you wish to use your own keys. Either way, it’s a pretty simple procedure and when you are done, you have link.
Before I started testing the speed of this equipment, though, I figured it was wise to benchmark the existing setup. I created an empty 1GB file to use for testing. 1GB might seem excessive, but I wanted to make sure I had enough time to get past TCP’s slow start and get a good idea of the maximum speeds involved. So the first set of benchmark numbers are sending data from my MacBook Air in the Office connected via 802.11g to Corinna’s Powerbook G4 in the Living Room connected via 802.11g to the second AP:
milgrim$ time nc -v -v -n pbook 2222 < big-file.bin
Connection to pbook 2222 port [tcp/*] succeeded!
968.65 real 9.04 user 44.13 sys
(1,073,741,824 bytes) / (968.65 seconds) = 8.45713106 Mbps
So the performance of my existing setup using WDS was so much worse than I had imagined. I had never really bothered to test it before, and was truly surprised at how slow it was. Next up, I plugged in the first Powerline adapter in the office and patched it into the WRT54G’s switch. I set up the second Powerline adapter in the Living Room and connected the Powerbook directly to it:
milgrim$ time nc -v -v -n pbook 2222 < big-file.bin
Connection to pbook 2222 port [tcp/*] succeeded!
545.85 real 7.88 user 65.73 sys
(1,073,741,824 bytes) / (545.85 seconds) = 15.007786 Mbps
For those keeping score at home, 15Mbps is much slower than 200Mbps. Even though I had never expected to get anywhere close to the advertised bandwidth, I was pretty disappointed. While 15Mbps was almost twice as fast as my existing setup, it was still rather slow. I tried removing some equipment from the equation, connecting the MacBook Air directly to the office powerline adapter (so the two laptops were, in theory, “directly connected”), but I got very similar numbers. I then wired both laptops directly into the same 100Mbps switch and re-ran the test, and got around 60Mbps, so the laptops themselves were not the limiting factor. I also tried placing the two endpoints in the same room, using different power outlets on the same wall, and the speed went up a bit closer to 25Mbps, although I managed to lose the hard numbers.
So where do I stand on this technology? Disappointed, but likely sticking with it for the time being. Even though the throughput is MUCH lower than I had hoped, it is still better than what I had. And until I can figure out a way to get real Cat5 between the two rooms, I don’t really see any other solution on the horizon. I’m certain that the slow speed is due to some noise in my wiring, but being that the wiring is new within the past 5 years, I’m guessing it is some device creating the noise. I guess I could try to figure out what it is, but I probably won’t. I’ll likely keep dicking around with the devices to see if I can get them to run any faster, and until then I’ll take what I can get. Someday I’ll either get physical wire between the two rooms or find another solution, and then I’ll unload the equipment.. But until then I guess it will have to do.