Powerline Networking... reviews are hard to find!

[Evadman]

I tried 2 of the netgear 85 Mbps units at my house when newegg had them on the cheap, and got maybe 1 or 2 mbps. I tried them in all different plugs, on the same breaker or different, same bus or different and I get crap on all of them.

I tried them at a friends house, (who by the way has the EXACT same house model, except mirrored and built 2 years after mine) and he got about 40 Mbps. Power line is something that depends on the wiring of our house a great deal, so YMMV.

I went wireless in the short term (Linksys Wireless N with 2 extra access points) while I run cat-5 in the house. Even with wireless N, I have a problem streaming movies to my media player.

 

[Eug]

The new 200 Mbps AV units are supposed to be better than the 85 Mbps units, in terms of handling noise. However, if you use surge suppressors, the performance will be terrible regardless. Were you using surge suppressors?

 

[Toadster]

The new 200 Mbps AV units are supposed to be better than the 85 Mbps units, in terms of handling noise. However, if you use surge suppressors, the performance will be terrible regardless. Were you using surge suppressors?

My setup has the units plugged directly into the wall socket (as prescribed by the manufacturer) - I agree that I'm not going to see anything near 150Mbps as the picture states. The speeds change throughout the day and during different load points on the house - whether the refridgerator is running, or the heat is on - it does fluctuate.

But it's still MUCH faster than my 802.11g connection at "FAIR" in the farthest bedroom. Homes are somewhat 'anti-wifi' with all the wood/sheetrock/doors in between everything. So for now I'm rather happy with the speeds.

 

[JackMDS]

Can you please tell us what the sustained speed is between the main feed entry (your main workstation) and the "lowest" points (Network wise) when transferring a file of about 100MB (in a MByte/sec. number) ?

 

[Eug]

My setup has the units plugged directly into the wall socket (as prescribed by the manufacturer) - I agree that I'm not going to see anything near 150Mbps as the picture states. The speeds change throughout the day and during different load points on the house - whether the refridgerator is running, or the heat is on - it does fluctuate.

Yeah, I have an IP camera out in the cold 200 feet from my house. It's too far for reliable wireless (and the IP camera I bought doesn't have wireless anyway), but there is a plug there so I put in powerline ethernet. I had tested it before several times for just surfing and light streaming and it worked fine, but I wasn't sure how consistent it'd be, given the distance. Well, with the IP camera it works perfectly, no matter how much I test it.

Luckily the max speed required with the camera is less than 5 Mbps, but then again the plug is 200 feet from my house and has garden lights running off it too.

I'll probably eventually run CAT5e or CAT6 there, but I've been holding off since it's the middle of winter and because I'd need gel-filled (flooded) outdoor cable which is expensive and harder to find. Maybe I'll wait until it isn't cold, or until when decent outdoor HD H.264 cameras are not too expensive cuz with this SD camera, powerline works perfectly fine.

But it's still MUCH faster than my 802.11g connection at "FAIR" in the farthest bedroom. Homes are somewhat 'anti-wifi' with all the wood/sheetrock/doors in between everything. So for now I'm rather happy with the speeds.

If you think that's bad, try cement walls.

Can you please tell us what the sustained speed is between the main feed entry (your main workstation) and the "lowest" points (Network wise) when transferring a file of about 100MB (in a MByte/sec. number) ?

I tried in various spots in the house. In the same room on the same circuit it worked well, but that doesn't count. Across the house in some spots I was getting as low as maybe 1-2 MB/s, but in other spots I was getting maybe 4-5 MB/s.

I'd try the one that's outside 200 feet from my house, but I don't want to bring my MacBook Pro out into the snow.

 

[JackMDS]

Thanks Eug that is a nice start for a frame of reference.:thumbsup:

I hope that more people can chime in, get off the Compu-Socio-Politico jargon, and and post more numbers.

.

 

[Eug]

So, just for you, I trudged through 10 cm deep snow to the Powerline adapter unit outside 200 feet away, with MacBook Pro in hand.

iMac --> Gigabit switch --> Gigabit switch (on other side of house, connected via CAT5e) --> 100 Mbps switch --> Powerline adapter (in house) ----------> Powerline adapter (outside) --> 100 Mbps switch --> MacBook Pro.

iperf tells me I get 9.5 to 10.4 Mbps. Transferring a hidef movie trailer file, I get just under 1.2 MB/s, or 9.5 Mbps.

Inside the house, these are my measurements with iperf:

In another room: 16.6 Mbps
In another room further away: 12.0 Mbps
At the opposite end of the house, through two circuit breaker panels: 0.6 Mbps

I have gotten as high as 35 Mbps through two circuit breaker panels, but the speed could drop to around 10 Mbps once in a while (microwave or dryer? not sure), and it only worked with one very specific plug. Another plug would consistently give me below 10 Mbps (sometimes more like 3 Mbps), even though it was in the same room as one that gave me 35 Mbps. The latter was a brand new dedicated circuit with no electronic devices on it though. The slower plug was an older circuit with all sorts of crap on it.

Note that this is not with Powerline AV. All my adapters are Netgear Powerline HD. It's the same speed as AV, but the chipset is different, and lost out for the standardization. (They were two competing but similar chipsets. AV won, mine didn't.)

Benchmarks of Powerline HD vs Powerline AV.

 

[Toadster]

iperf ?

need something windows based...

 

[Eug]

iperf runs on Windows

 

[Toadster]

iperf runs on Windows

Thanks!

ran just the default test

------------------------------------------------------------
Client connecting to 192.168.1.74, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[148] local 192.168.1.64 port 49214 connected with 192.168.1.74 port 5001
[ ID] Interval Transfer Bandwidth
[148] 0.0-10.0 sec 20.4 MBytes 17.1 Mbits/sec

and from the other end:

Server listening on TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
------------------------------------------------------------
Client connecting to 192.168.1.64, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[160] local 192.168.1.74 port 64887 connected with 192.168.1.64 port 5001
[184] local 192.168.1.74 port 5001 connected with 192.168.1.64 port 49367
[ ID] Interval Transfer Bandwidth
[160] 0.0-10.0 sec 14.4 MBytes 12.1 Mbits/sec
[184] 0.0-10.0 sec 14.3 MBytes 12.0 Mbits/sec

and with bi-directional in serial fashion:
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
------------------------------------------------------------
Client connecting to 192.168.1.64, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[160] local 192.168.1.74 port 65083 connected with 192.168.1.64 port 5001
[ ID] Interval Transfer Bandwidth
[160] 0.0-10.0 sec 21.1 MBytes 17.7 Mbits/sec
[132] local 192.168.1.74 port 5001 connected with 192.168.1.64 port 49377
[ ID] Interval Transfer Bandwidth
[132] 0.0-10.0 sec 19.8 MBytes 16.6 Mbits/sec

 

[Toadster]

interesting - at 4 to 8 threads, I can max the line around 34.8 Mbits/sec.

iperf -c 192.168.1.64 -P 8
------------------------------------------------------------
Client connecting to 192.168.1.64, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[208] local 192.168.1.74 port 51252 connected with 192.168.1.64 port 5001
[200] local 192.168.1.74 port 51251 connected with 192.168.1.64 port 5001
[192] local 192.168.1.74 port 51250 connected with 192.168.1.64 port 5001
[184] local 192.168.1.74 port 51249 connected with 192.168.1.64 port 5001
[176] local 192.168.1.74 port 51248 connected with 192.168.1.64 port 5001
[152] local 192.168.1.74 port 51245 connected with 192.168.1.64 port 5001
[168] local 192.168.1.74 port 51247 connected with 192.168.1.64 port 5001
[160] local 192.168.1.74 port 51246 connected with 192.168.1.64 port 5001
[ ID] Interval Transfer Bandwidth
[160] 0.0-10.0 sec 5.24 MBytes 4.39 Mbits/sec
[192] 0.0-10.0 sec 5.24 MBytes 4.39 Mbits/sec
[168] 0.0-10.0 sec 5.19 MBytes 4.34 Mbits/sec
[200] 0.0-10.0 sec 5.25 MBytes 4.39 Mbits/sec
[176] 0.0-10.0 sec 5.20 MBytes 4.35 Mbits/sec
[184] 0.0-10.0 sec 5.16 MBytes 4.32 Mbits/sec
[208] 0.0-10.0 sec 5.11 MBytes 4.27 Mbits/sec
[152] 0.0-10.0 sec 5.20 MBytes 4.35 Mbits/sec
[SUM] 0.0-10.0 sec 41.6 MBytes 34.8 Mbits/sec

 

[Emulex]

interesting - at 4 to 8 threads, I can max the line around 34.8 Mbits/sec.

iperf -c 192.168.1.64 -P 8
------------------------------------------------------------
Client connecting to 192.168.1.64, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[208] local 192.168.1.74 port 51252 connected with 192.168.1.64 port 5001
[200] local 192.168.1.74 port 51251 connected with 192.168.1.64 port 5001
[192] local 192.168.1.74 port 51250 connected with 192.168.1.64 port 5001
[184] local 192.168.1.74 port 51249 connected with 192.168.1.64 port 5001
[176] local 192.168.1.74 port 51248 connected with 192.168.1.64 port 5001
[152] local 192.168.1.74 port 51245 connected with 192.168.1.64 port 5001
[168] local 192.168.1.74 port 51247 connected with 192.168.1.64 port 5001
[160] local 192.168.1.74 port 51246 connected with 192.168.1.64 port 5001
[ ID] Interval Transfer Bandwidth
[160] 0.0-10.0 sec 5.24 MBytes 4.39 Mbits/sec
[192] 0.0-10.0 sec 5.24 MBytes 4.39 Mbits/sec
[168] 0.0-10.0 sec 5.19 MBytes 4.34 Mbits/sec
[200] 0.0-10.0 sec 5.25 MBytes 4.39 Mbits/sec
[176] 0.0-10.0 sec 5.20 MBytes 4.35 Mbits/sec
[184] 0.0-10.0 sec 5.16 MBytes 4.32 Mbits/sec
[208] 0.0-10.0 sec 5.11 MBytes 4.27 Mbits/sec
[152] 0.0-10.0 sec 5.20 MBytes 4.35 Mbits/sec
[SUM] 0.0-10.0 sec 41.6 MBytes 34.8 Mbits/sec

yes its because tcp doesn't like that much packet loss. the nagle/autotuning freaks out.

try a udp test you'll see what i mean. these things really do run 90-100% packet loss over udp.

I've seen one of these drop down to 50% packet loss on a perfect circuit with udp. thats 100mbit. (plugged ass2ass sorta).

 

[Eug]

What are you using to measure packet loss?

 

[Toadster]

yes its because tcp doesn't like that much packet loss. the nagle/autotuning freaks out.

try a udp test you'll see what i mean. these things really do run 90-100% packet loss over udp.

I've seen one of these drop down to 50% packet loss on a perfect circuit with udp. thats 100mbit. (plugged ass2ass sorta).

yes UDP is about 1.10Mbps - much reduced

here's a test from a PC that is Wireless based (close to the .64 host and AP)

iperf -c 192.168.1.64
------------------------------------------------------------
Client connecting to 192.168.1.64, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[148] local 192.168.1.36 port 52154 connected with 192.168.1.64 port 5001
[ ID] Interval Transfer Bandwidth
[148] 0.0-10.0 sec 23.3 MBytes 19.5 Mbits/sec

iperf -c 192.168.1.64 -r
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
------------------------------------------------------------
Client connecting to 192.168.1.64, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[160] local 192.168.1.36 port 52239 connected with 192.168.1.64 port 5001
[ ID] Interval Transfer Bandwidth
[160] 0.0-10.0 sec 24.1 MBytes 20.2 Mbits/sec
[128] local 192.168.1.36 port 5001 connected with 192.168.1.64 port 63126
[ ID] Interval Transfer Bandwidth
[128] 0.0-10.0 sec 24.5 MBytes 20.6 Mbits/sec

 

[Toadster]

and just for fun - here's my ISP bandwidth test from a PC on PLE Networking



Uploaded with ImageShack.us

 

[Eug]

I'd still be interested in what people are using to measure packet loss.

As for the ISP bandwidth test, as you're likely aware, you may be limited by the ISP service's speed caps.

Anyways, this does illustrate the utility of powerline networking:

1) Usually great for basic internet surfing, and often can easily max out the speed for average consumer broadband.
2) Often quite OK for basic networking around the house. Roughly 802.11g type speeds, but usually with better reliability. IOW, one should expect 10 Base-T speeds, but sometimes one can also get speeds several times real-world average 10 Base-T speeds.
3) SD streaming of downloaded video should usually work over powerline. A standard 45 minute TV show file is a piece of cake, since it only needs maybe 1-2 Mbps.
4) HD streaming is often very problematic. Don't expect to get 10 Mbps HD H.264 to stream 100% consistently, even if your benchmarks are saying you can get 20 Mbps over powerline. In some setups it can be consistently that high, but IMO it's more the exception than the rule.

I'll probably lay Gigabit Ethernet in the backyard eventually, but right now my 9+ Mbps over powerline is more than sufficient. However, the reason is that my deer cam is only a 640x480 camera. It is set at 2 Mbps currently, and it worked fine too with the maximum 4.5 Mbps setting. However, going forward, powerline isn't going to be sufficient if I ever upgrade the camera. Current 1.3 Megapixel (1280x1024) cameras max out around 8 Mbps. My 9 Mbps line to the backyard just doesn't have enough breathing room to support that. And who knows what consumer 1080p IP cams will be? 15 Mbps?

 

[Toadster]

What are you using to measure packet loss?

nothing currently - I mean, I just found out that iperf works LOL

UDP traffic is meager at best so the medium is very lossy...
I think you nailed it - it has 802.11g speeds but reliable.

I have streamed HD content from my NAS to my Boxee which goes Powerline - Router - WLAN - MacMini running Boxee - so lots of kinks to break...

we only had one movie where Boxee had to re-cache a few times - otherwise it's flawless.

sure if I could run Gigabit around the house it would be optimal - but considering I've spent under $200 and solved the immediate problem - that is pretty cost efficient too.

oh - and any of you wanting to reply about how cheap CAT6 cable is and I could get a spool and run it inside/outside/through my house - you're welcome to come over and share some beers while I watch you chase cable

 

[Emulex]

you login to the device itself lol. they all have web consoles where you can change the key and watch the throughput etc. most are the same chipset.

my main biggie is the lack of warranty of goods for damage due to lightning.

APC warrants my goods. These guys do not. that is not a good thing. everything is hardwired here. a lightning strike could easily nuke a switch and few pc's.

 

[TexasEd]

Just installed the Netgear Powerline AV series in my house. I got the version with a 4 port hub at the remote end.

My house was built in 2005
I'm hooking it up to a AVR for Internet radio and a PS3 for PSN, Netflix, etc.

It works great for the Internet radio, PSN and youtube on PS3. I have not created my netflix account yet so have not tested that.

I need to run the throughput tests but it does not seem laggy and the Youtube videos load fast and do not pause. Hookup was as easy as could be.

What should I use for throughput testing to compare apples to apples with other numbers?

I imagine something on my wired desktop through the netgear powerline to a wired laptop (with PS3 and AVR disconnected) would be a good way to test the throughput on just this portion of the network.

Also how does the lockout feature/button on this work?

 

[YBS1]

I have a few of the Belkin 200Mbps adapters hooked up to my network. They actually perform pretty well here (as mentioned this is a YMMV depending upon your house), I usually get somewhere in the range of 3-5MBps for transfers, though I've only ever had a blu-ray player hooked up to the one farthest from the router so I have no idea how well it holds up there. One thing that I've not seen mentioned here and I don't know if it's typical or specific to my network, they carry a considerable latency penalty. 4-10ms to the router if memory serves me correctly. Just an FYI in case someone is considering them for a gaming connection (adding 10ms before you even get off your home network not really a great thing).

 

[Toadster]

you login to the device itself lol. they all have web consoles where you can change the key and watch the throughput etc. most are the same chipset.

my main biggie is the lack of warranty of goods for damage due to lightning.

APC warrants my goods. These guys do not. that is not a good thing. everything is hardwired here. a lightning strike could easily nuke a switch and few pc's.

I've done some searches but can't find a direct way to login to the device itself - can you inform?

You bring up a good point about no power surge supression - I do have my surge strip plugged into the Netgear itself - so yes, while I may lose the networking device - the rest of my stuff should be OK.

maybe a good time to get one of those whole-house surge arrestors that can be installed by the power company?

 

[Eug]

One thing that I've not seen mentioned here and I don't know if it's typical or specific to my network, they carry a considerable latency penalty. 4-10ms to the router if memory serves me correctly. Just an FYI in case someone is considering them for a gaming connection (adding 10ms before you even get off your home network not really a great thing).

Well, I actually thought that was pretty good, because it's significantly less lagginess (and better reliability) than I had experienced with secured wireless way back when. I get pings of about 2-11 ms for powerline.

But yeah, wired Ethernet is obviously going to much better. I get < 1 ms over wired, even through multiple switches.

 

[Toadster]

2-11ms is pretty good if you ask me, most people don't notice under 30ms for gaming