Powerline Ethernet Adapters
First, we all know that wired ethernet is best. If the bandwidth specified isn’t seen, there are other issues. It isn’t the wires.
However, here are times when some other network method is needed. Typically, that is wifi, but we’ve learned over the years that wifi sucks too. The advertised connection rates do not reflect real-world bandwidth. Getting even half that amount is lucky. We’ve learned to live with it, since for most people anything over 10 Mbps is fine.
Powerline is about the same as wifi based on my testing today. Picked up a set of ZyXEL 600 Mbps Mini Powerline AV2 Gigabit Adapter this week to extend a wired ethernet connection from the wiring closet upstairs to the opposite part of the house, the den, downstairs. Primary use will be to feed a cheap Raspberry Pi media player.
The results – 60 Mbps. Good enough, but hardly worth paying more than that for a 500Mbps model. That is just 10% of the bandwidth promised on the box!
Rant About 10% of Advertised
Why is it ok to get just 10%? Seems like false advertising – if your car got 10% of the advertised MPG, they’d be sued. If you bought a box of corn flakes and it was only 10% of the advertised amount, you’d never buy it again.
Comcast, AT&T, Verizon, Sprint all claim “up to XYZ Mbps” bandwidth. They usually do better than 10% of the advertised rate, but not always (I hear that from DSL customers). My comcast connection is almost always within 5-10% of the advertised bandwidth – sometimes it is higher! I’ve never heard of any DSL customer getting more bandwidth than they paid for – not once.
Why do we put up with this for network performance?
The Numbers
Ran a bandwidth testing tool to see the bandwidth with the two ZyXEL 600 Mbps devices. Engineer here. Feelings mean nothing – show me the numbers.
It works, but is very far off the advertised bandwidth. Don’t really see any reason to pay extra for 600 Mbps over 500 Mbps versions. The 1200 Mbps versions must be a joke. Read the articles over at TheWirecutter or SmallNetBuilder – any websearch will find them. I don’t provide the links here on purpose.
For the details:
Baseline – using same server, same client, and same network hops. There are two $20 GigE switches between the client and the server in this test. Just the last link will change, from the second switch to the client netbook.
- over a GigE wired connection: 921 Mbits/sec
Scenario A: The first setup is from the hallway to a bedroom. The same network devices and computers were used for both. Just the last hop from the switch to the client netbook was swapped out. The physical distance between the two adapters was about 8 ft, but I suspect they were on different house circuits.
- over these devices: 219 Mbits/sec
Better than wifi has ever been in this house.
Scenario B: Left everything else the same, but moved the bedroom adapter downstairs and took the netbook with me.
- 60.0 Mbits/sec (ran this test 5 times – it was always between 59.0 and 61 Mbps)
Conclusions:
- Distance matters greatly. The claimed 600 Mbps is marketing only. It is a lie. All the reading I’ve done shows that NOBODY gets even 300Mbps.
- Getting 200 Mbps won’t happen in any real-world case unless the router/switch is either in the same room or on the same circuit. Doubtful anyone will deploy these for that purpose.
- The upstairs/downstairs connections loose about 140 Mbps. Ouch. Plus that test doesn’t even reflect the planned deployment location in the wiring closet – which is farther.
- 60 Mbps
- 60.7 Mbits/sec – 2nd run a few min later
- For normal household uses like internet gaming and streaming hidef videos, 60 Mbps is fine. For video encoding or transferring large files over this part of the network, it is less bandwidth than any 100 base-T connection. 2 bluray streams over this network would likely show stuttering. However, 1 bluray and 3 DVD streams would be fine.
- Wired ethernet is ALWAYS better than wifi or powerline. Use that whenever possible. Ethernet speeds do not drop off over normal house distances – at all. I have a CAT5e cable that we bring out to make these same connections from upstairs to downstairs (ugly cable) and it tests over 920 Mbps.
RAW Numbers:
The raw numbers from iperf – hadar is a Core i7 “server”, GigE connected, doing nothing else. The client is an Acer C720 Chromebook running Ubuntu Server GigE connected through a USB3-to-GigE adapter. It isn’t running any other network tasks. House was build in 1994 with typical copper wiring and breaker in the garage.
Scenario A Data:
$ iperf \-c hadar
-————————————————————————————-
Client connecting to hadar, TCP port 5001
TCP window size: 43.8 KByte (default)
-————————————————————————————-
[ 3] local 172.22.22.13 port 58614 connected with 172.22.22.6 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 262 MBytes 219 Mbits/sec$ iperf \-c hadar
-————————————————————————————-
Client connecting to hadar, TCP port 5001
TCP window size: 43.8 KByte (default)
-————————————————————————————-
[ 3] local 172.22.22.13 port 58632 connected with 172.22.22.6 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 1.07 GBytes 921 Mbits/sec
Scenario B Data:
$ iperf \-c hadar
-————————————————————————————-
Client connecting to hadar, TCP port 5001
TCP window size: 43.8 KByte (default)
-————————————————————————————-
[ 3] local 172.22.22.13 port 58791 connected with 172.22.22.6 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.1 sec 72.0 MBytes 60.0 Mbits/sec
Conclusion
For normal household uses like internet gaming and streaming a few hidef videos, 60 Mbps is fine.
For video encoding over network storage or transferring large files over this part of the network, it is 40% less bandwidth than any wired 100-base-T connection. 2 bluray streams over this network would likely show stuttering. However, 1 bluray and 3 DVD streams would be fine. Also, 4 netflix hidef streams should be find.