Would Powerline ethernet make my speeds faster for my SAMBA server?

[JQLeitch]

I currently have a very low-cost, but effective NAS solution which is a Raspberry Pi using SAMBA and a external hard drive formatted in EXT4, does everything fine in terms of watching blu rays over wireless.

The thing is, is there's a few things that it is useless in and that's uncompressed videos, such as uncompressed trailers (e.g. 300MB file for a minute)

I use wireless and this is when I can't view uncompressed trailers without a few buffers but over ethernet it's flawless! My write speeds from PC to the NAS are:

Wireless: 2-4MB/s
Wired: 8.9-12MB/s

The thing is, is I have no way to use Ethernet without running a long cable down my staircase, which my housemates would not like. Would powerline ethernet give me any of the performance that Wired does, at least comparable and better than Wireless?

 

[JackMDS]

When it comes to Powerline there is No way to know, you have to try.

 

[azazel1024]

Most likely, but as Jack said, no way to know without trying it.

Considering cost, I'd look at upgrading the wireless end of things. No way of knowing what you are using, distances, interference, etc...but its either a darned long distance through a lot of walls, or older 11n wireless intead of a newer wireless setup.

At any rate, with AV500/600 powerline adapters, figure anything from about 30Mbps to around 200Mbps, just depending on the idealness of the setup, your electrical wiring and what is plugged in to that wiring. Figure most likely in the 40-60Mbps range. So...yeah, somewhat better performance, but you are unlikely to get in to the 9-12MB/sec range.

Same room, or same circuit with little else plugged in (especially wall warts/AC-DC transformers) and you might get >100Mbps, but not likely with typical setups/homes.

 

[JQLeitch]

Seems @azazel1024 was right. I bought a pair of powerline ethernet devices from TP-Link, which were cheap and my speeds are very near Ethernet, averaging around 8MB/s but fluctuating far above. Very happy with my purchase xD

 

[Eug]

Just hope your wife doesn't use a microwave oven and your daughter doesn't use her hairdryer at the same time as when you're trying to stream your trailers.

I had this problem.

I could usually stream HD over Powerline when I was alone, but when my wife was in the house, sometimes that connection got very erratic. She didn't have to be in any of the same rooms. Likely just the fact that she was using electrical devices in the house added enough additional noise to the system that screwed up some of my powerline signals.

It sounds like you were lucky, but don't be completely surprised sometime in the future if you find that 8 MB/s minimum speed is not always the minimum speed.

 

[azazel1024]

Just hope your wife doesn't use a microwave oven and your daughter doesn't use her hairdryer at the same time as when you're trying to stream your trailers.

I had this problem.

I could usually stream HD over Powerline when I was alone, but when my wife was in the house, sometimes that connection got very erratic. She didn't have to be in any of the same rooms. Likely just the fact that she was using electrical devices in the house added enough additional noise to the system that screwed up some of my powerline signals.

It sounds like you were lucky, but don't be completely surprised sometime in the future if you find that 8 MB/s minimum speed is not always the minimum speed.

Just hope she doesn't use a microwave and you have both a 2.4GHz wireless link AND powerline on the same connection. Then you are really screwed.

I actually never bothered to test until recently, but having my microwave on has very little 2.4GHz impact...UNLESS my client is in the kitchen. If in the kitchen I see an immediate 20-30% drop in 2.4GHz throughput until the microwave stops. Just about anywhere else in the house and the hit is tiny if even noticable (~5% hit). Same with my 2.4GHz portable phone. Using it causes almost no impact, unless near the client (like talking on the phone and sitting on my laptop), then I see around a 10-15% hit, sometimes. Only when the phone is in actual use.

 

[Eug]

Yeah for 2.4 GHz wifi the big drop is when the the microwave and client is in the same room. Otherwise I don't really notice it. I guess it could be a problem though if the router itself was in the same room as the microwave or else the microwave was in between the router in the client.

 

[azazel1024]

Yeah for 2.4 GHz wifi the big drop is when the the microwave and client is in the same room. Otherwise I don't really notice it. I guess it could be a problem though if the router itself was in the same room as the microwave or else the microwave was in between the router in the client.

I guess I need to take my router off my kitchen counter microwave...huh?



That said, I HAVE seen a friend who had their router sitting on a counter top microwave in their apartment. Couldn't figure out why their wifi went to hell every now and again...

 

[Emulex]

when the dyson is not activated (100% packet loss) I found powerline ethernet at short distances good for 90% packet loss at rated mbps. IE 200 megabits => 20 megabits actual performance at point blank range.

So yeah, it sucks and a small cheapo brick charger can take that 20 megabits actual and make it ZERO.

 

[PliotronX]

The Tplinks are great, I see 80 to 120Mbps and no packet loss even when the laser printer is flickering lights. So much more consistent than wireless but ultimately I'm going to run GigE as I want more speed though the internet maxes at 40Mbps here so it's not a priority. I'd give it a shot before upgrading expensive wireless devices.

Oh hehe you got em, nice. Yeah we need AV3 MiMO and MoCA advancements not 300 dollar routers that still share airwaves with douchebags that leave their routers on auto and keep hopping to your channel.

 

[azazel1024]

when the dyson is not activated (100% packet loss) I found powerline ethernet at short distances good for 90% packet loss at rated mbps. IE 200 megabits => 20 megabits actual performance at point blank range.

So yeah, it sucks and a small cheapo brick charger can take that 20 megabits actual and make it ZERO.

Wireshark it, that isn't 90% packet loss. It is dropping modulation rate significantly to compensate with the SINR environment it is working in. Same thing happens with wlress. If you had 90% packet loss you'd probably only be getting 1-2Mbps as you'd have so many resends and lost ACK packets it would make the connection nearly unworkable (maybe entirely unworkable).

I agree that better powerline and MoCA would be nice, but you have limits there. MoCA is typically pretty clean, but you are dealing with two wires. Ethernet you have 8 wires...which means 4 times the possible speed dealing with similar technologies (and it also is cleaner than coax, in general, at least for the better Cat wiring standards). So MoCA will never be able to rival ethernet. Powerline is worse because it is an extremely noise environment and at best you have 3 wires, 2 of which are bonded at some point, which ups the difficulty and they run parallel to each other. Powerline is a lot noisier than wireless is most of the time, the only advantage it really has over wireless is that walls don't attenuate a powerline signal, but they do a wireless signal.

Considering that most people use wireless and wireless is easier to implement, I think we will continue to see wireless outpace MoCA and powerline.

 

[PliotronX]

Wireshark it, that isn't 90% packet loss. It is dropping modulation rate significantly to compensate with the SINR environment it is working in. Same thing happens with wlress. If you had 90% packet loss you'd probably only be getting 1-2Mbps as you'd have so many resends and lost ACK packets it would make the connection nearly unworkable (maybe entirely unworkable).

I agree that better powerline and MoCA would be nice, but you have limits there. MoCA is typically pretty clean, but you are dealing with two wires. Ethernet you have 8 wires...which means 4 times the possible speed dealing with similar technologies (and it also is cleaner than coax, in general, at least for the better Cat wiring standards). So MoCA will never be able to rival ethernet. Powerline is worse because it is an extremely noise environment and at best you have 3 wires, 2 of which are bonded at some point, which ups the difficulty and they run parallel to each other. Powerline is a lot noisier than wireless is most of the time, the only advantage it really has over wireless is that walls don't attenuate a powerline signal, but they do a wireless signal.

Considering that most people use wireless and wireless is easier to implement, I think we will continue to see wireless outpace MoCA and powerline.

Wireless isn't noisy? When you have 20 neighboring AP's all on the same channels, I don't care what router it is, it is going to be slower than a powerline adapter. Happy wireless users tend to be secluded and connected within arms reach of the AP but in more dense areas, in the real world, wireless is awful. Inconsistent pings and throughput, susceptible to interference and driveby hacking. I typically only recommend wireless for mobile devices. There is no way in hell home Wi-Fi has more potential than any copper standard. All that work in wireless research goes down the toilet through obstacles and interference.

 

[imagoon]

I agree that better powerline and MoCA would be nice, but you have limits there. MoCA is typically pretty clean, but you are dealing with two wires. Ethernet you have 8 wires...which means 4 times the possible speed dealing with similar technologies

This very incorrect. Coax and TwinAX significantly out performs 4 twisted pairs, assuming same core gauge as the twisted pair.

There is a reason why 40GBASE-CR4 and 100GBASE-CR10 exist in the real world (as in you can buy it now, today) and 40GBASE-T is over a nonexistent Cat8 and 100GBASE-T is undefined.

Coax is able to shield out noise and contain the signal in the cable than twisted pair.

 

[azazel1024]

This very incorrect. Coax and TwinAX significantly out performs 4 twisted pairs, assuming same core gauge as the twisted pair.

There is a reason why 40GBASE-CR4 and 100GBASE-CR10 exist in the real world (as in you can buy it now, today) and 40GBASE-T is over a nonexistent Cat8 and 100GBASE-T is undefined.

Coax is able to shield out noise and contain the signal in the cable than twisted pair.

At long ranges, sure. At short distances, no.

Cat6a is supposedly rated for 40GBase-t to 30 meters, even though that is also what Cat8 is supposed to be. There is no coax for 40GBase-t or 100Gbase-t, it is Twinaux only, which is basically 2 wire pairs, compared to 4 wire pairs in cat5+ cabling.

Yes, the shielding is quite good and the gauge is typically thicker, however, most 6a is also shielded (to a pretty good standard) and Cat7 and the theoretical 8 all is shielded, and have MUCH longer distances than direct attached twinaux for 40/100GbE.

You have more channels to populate with 4 wire pairs than you do with a single wire pair. The performance over distance might be better, but with 4 times as many "channels" to populate, you can cram a lot more data down that pipe until you reach a distance at which you lose the signal.

Also, yes, wireless can be noisy, but in relatively good settings (which aren't that rare) it is a HELL of a lot less noisy than powerline in. Sure, a lot of people live where they have 20 overlapping networks, but a lot (a lot a lot) of people don't and at most might have a couple of network to deal with, or even possibly none at all. Powerline, you have gobs of noise on the wire and really terrible attenuation charateristics.

 

[imagoon]

At long ranges, sure. At short distances, no.

Cat6a is supposedly rated for 40GBase-t to 30 meters, even though that is also what Cat8 is supposed to be. There is no coax for 40GBase-t or 100Gbase-t, it is Twinaux only, which is basically 2 wire pairs, compared to 4 wire pairs in cat5+ cabling.

Yes, the shielding is quite good and the gauge is typically thicker, however, most 6a is also shielded (to a pretty good standard) and Cat7 and the theoretical 8 all is shielded, and have MUCH longer distances than direct attached twinaux for 40/100GbE.

You have more channels to populate with 4 wire pairs than you do with a single wire pair. The performance over distance might be better, but with 4 times as many "channels" to populate, you can cram a lot more data down that pipe until you reach a distance at which you lose the signal.

Twinax is one wire pair. I am also pretty sure there is no 40Gbps on CAT6A "spec" other than "pray it works."

Distance is also irrelevant, for any distance, current designs show coax "wins." 4 twisted pairs, even shielded suffers from parallelism issues.

This is similar to SATA and PATA. 80 wires in a PATA cable "should" be a ton faster but it isn't due to timing overhead. This is the issue that 40Gbps and 100Gbps is seeing. This is also the main reason that 10GB-T barely exists in the twisted pair world so far. It is cheaper and more reliable to throw twinax and fiber at the problem.

 

[azazel1024]

No, Twinax isn't. It has two conductors plus ground (the shield). Cat cabling (well, 5+) is 4 pairs of conductors and grounds.

10GBase-T doesn't barely exist in the twisted pair world. Cat5e can carry it up to 45 meters, 6 to 55 meter and 6a to 100 meters. Twinax can carry 10GbE to an absolute maximum of 10 meters.

That is a huge difference in how far.

40GbE can work on cat6a. This issue is is that Cat6a isn't tested above 1000MHz where 40GbE lives. Tests show that it can work over short distances. 7a is "where its at" for long distance (IIRC 7a is another name for 8, problem with two standards bodies determining things, they call things different names until someone decides what the common use name is) as it can take 40GbE to 100 meters in theory and 100GbE possibly to 30 meters.

Cat 7 (or maybe it is 7a) has the individual wires shielded, which relieves the cross talk issues, which makes it massively superiod to twinax (all wires shielded, and a heck of a lot more wires. Twinax also has parellelism issues, as the two conductors are not shielded from each other, only from alien crosstalk by the shielded ground jacket).

The perk of coax is that it has a very thick conductor and drain compared to cat wiring. Typical THICK cat wiring is only 23 gauge. Coax (RG6) is 18 gauge and Twinax might be slightly thicker (not sure). A 1000MHz signal disipates relatively quickly on 23 gauge, but on 18 gauge you have much better attenuation characteristics. The shield is a nice to have for environmental noise filtering.

When it comes down to it, cat wiring, unless it is shielded, has worse noise characteristics from cross talk and alien cross talk and also EMI than coax does. It also has worse attenuation. However, it has significantly more wires to talk across. In a home environment, that coax is not clean and you have many more issues, such as tons and tons of hook-ups on the wires, splitters, etc which causes massive attenuation and noise compared to what cat cabling has to experience. So, MoCA has a lot to deal with compared to Cat cabling.

In a straight up one wire to one wire comparison, Coax as it is typically found (Twinax or RG6 single conductor coax) is much better over range than Cat cabling, because of the lower attenuation and susceptibility to cross talk and noise (the later only if you are comparing non-shielded, with shielded the susceptibility to noise and alien cross talk is vaguely the same, even if you have cross talk/parallelism).

This is why you can get a 1000m coax hook-up and run 100Mbps down it, when there is no Earthly way you can do that with any Cat cabling in existance. Get a short run though and I don't care how much you try, 4 wire pairs is going to beat out a single wire pair, or a dual conductor and single ground solution.

 

[imagoon]

So we are way off topic but from your explanation, I highly suggest you read up on parallel / multiple path data systems and timing issues. There are many times with "4 pairs" is worse than 1. From your argument we should have never gone from PATA to SATA. It was clear you didn't understand what I am talking about.

Also Twinax can go much farther than 10m. It is just at about 20-30m the cost of the cable and the power budget makes fiber much more attractive.

 

[Emulex]

There is powerline that uses multi-point transmitter/repeater kind of like 3-antennta/3 Transmitter wifi does! this can help reduce the 90%(UDP) packet loss that results in 10% rated performance using some sort of vectoring/Error-correction methods for powerline. It is not standard but IIRC someone sold it as a performance near-gigabit set of adapters (3 needed).

Good read using mino over PLC: http://www.marvell.com/wireline-networking/ghn/assets/MIMO-HGF.pdf

 

[azazel1024]

So we are way off topic but from your explanation, I highly suggest you read up on parallel / multiple path data systems and timing issues. There are many times with "4 pairs" is worse than 1. From your argument we should have never gone from PATA to SATA. It was clear you didn't understand what I am talking about.

Also Twinax can go much farther than 10m. It is just at about 20-30m the cost of the cable and the power budget makes fiber much more attractive.

The comparisons aren't equitable. There are also significant advantages in SATA over PATA from the fewer wire pairs. Space in a case being one of the big ones, as it is rather difficult to wire up a bunch of PATA drives comapred to SATA drives.

There is a reason why 10GBase-t is being pushed in to data centers over fiber and twinax, as it is becoming cheaper to deploy, and it DOES have much longer range than twinax.

 

[azazel1024]

There is powerline that uses multi-point transmitter/repeater kind of like 3-antennta/3 Transmitter wifi does! this can help reduce the 90%(UDP) packet loss that results in 10% rated performance using some sort of vectoring/Error-correction methods for powerline. It is not standard but IIRC someone sold it as a performance near-gigabit set of adapters (3 needed).

Good read using mino over PLC: http://www.marvell.com/wireline-networking/ghn/assets/MIMO-HGF.pdf

90% packet loss would make TCP unusable. I don't think you understand how powerline (and by extension) wifi works. What you have is a rated modulation spec, say AV500 for powerline. Then, what happens is the powerline adapters talk to each other to determine how favorable (or not) the signal and background noise are. They then agree upon a modulation scheme to use, which results in a connected modulation rate, this is what is displayed by the bridge, say 120Mbps. That isn't packet loss at all, that is just the best modulation rate the bridges can talk to each other and still reliably hear each other.

Then you have ACTUAL data rate, which is curtailed by forward error correction (on 11n it is about a 24% overhead to to FECC), ack packets (in the case of TCP), header overhead, protocol overhead and finally lost packets.

This is why you have an AV500 bridge pair, that might be at a 120Mbps link rate and you might only be getting 50Mbps actual through put. Same thing with Wifi, why your 300Mbps wireless connection might actually be connected at only 180Mbps and you might only be getting an actual 80Mbps on that connection.

Things get noisier, or you get quieter (or both), you have to talk slower to be understood.

There are a couple of MIMO adapters for powerline on the market right now. Improvements are minor (seemingly in the 20% range in most cases). Powerline is always going to be an extremely noisy environment and is at best going to have edge cases where it is better than wireless. That said, improvements are improvements. Where I could see powerline being a VERY nice thing is for IoT smart appliances. Why work on wireless, or why have that be the only option. If a smart plug, Fridge, radio, etc. could all talk a home hub over powerline instead of wifi, that would be better than cluttering up wifi that much more. Especially as most of those device's data rates are going to be very minimal.

 

[Eug]

There are a couple of MIMO adapters for powerline on the market right now. Improvements are minor (seemingly in the 20% range in most cases). Powerline is always going to be an extremely noisy environment and is at best going to have edge cases where it is better than wireless. That said, improvements are improvements. Where I could see powerline being a VERY nice thing is for IoT smart appliances. Why work on wireless, or why have that be the only option. If a smart plug, Fridge, radio, etc. could all talk a home hub over powerline instead of wifi, that would be better than cluttering up wifi that much more. Especially as most of those device's data rates are going to be very minimal.

I have some circuit pairs in my house where I get exactly a 0 Mbps data rate over powerline most of the time. WiFi works OK, at least with my last router purchase.

I have a bigger house than average, but that should be more of a disadvantage for WiFi than powerline. Perhaps the fact that I have a couple of electrical subpanels may be screwing stuff up. Whatever the case, if I had to choose one or the other, I'd choose WiFi. Having the option of both WiFi and powerline would be great though.