What is my strange home network setup missing?
November 30, 2008 2:27 PM Subscribe
Home networking question: What piece of hardware do I need to share a ... actually I don't know how to phrase it correctly so I've included links to 2 simple schematic drawings. I need help connecting a file server and an Xbox 360 to my home network.
Basically what I want is this. What should that mystery box be? Would a switch work? A network hub? Another router? The file server and the 360 don't exist yet, but I've checked the powerline network connection and it works. Wireless can be wonky in my house, and I don't want to connect the 360 wirelessly anyway. The router and cable modem (oh, there's an assumed cable modem between router and internet connection; sorry) can't be moved. Bizarre, I know, but true. Would this work? Are there other problems I haven't foreseen? Would server performance be hurt? I eventually want to set this server up to stream media, serve to me through a VPN connection when I'm elsewhere, etc.
If that setup wouldn't work, or if there's no such piece of hardware as I would need for the mystery box, would this setup work? Would the two powerline adapters interfere with each other?
Basically what I want is this. What should that mystery box be? Would a switch work? A network hub? Another router? The file server and the 360 don't exist yet, but I've checked the powerline network connection and it works. Wireless can be wonky in my house, and I don't want to connect the 360 wirelessly anyway. The router and cable modem (oh, there's an assumed cable modem between router and internet connection; sorry) can't be moved. Bizarre, I know, but true. Would this work? Are there other problems I haven't foreseen? Would server performance be hurt? I eventually want to set this server up to stream media, serve to me through a VPN connection when I'm elsewhere, etc.
If that setup wouldn't work, or if there's no such piece of hardware as I would need for the mystery box, would this setup work? Would the two powerline adapters interfere with each other?
Is there a reason the server couldn't live near your other computer?
posted by anarchivist at 2:51 PM on November 30, 2008
posted by anarchivist at 2:51 PM on November 30, 2008
Response by poster: Is there a reason the server couldn't live near your other computer?
Not a very good reason. It's really cluttered at that desk, and I'd rather have the server in the relatively clean area at the other end of the powerline adapter. But really that's a clutter problem and not a server problem. I should be taking care of that either way. Yeah, if the server were sufficiently small and quiet I might be able to locate it up there. Thanks for the suggestion. I hadn't even thought of it.
posted by penduluum at 3:10 PM on November 30, 2008
Not a very good reason. It's really cluttered at that desk, and I'd rather have the server in the relatively clean area at the other end of the powerline adapter. But really that's a clutter problem and not a server problem. I should be taking care of that either way. Yeah, if the server were sufficiently small and quiet I might be able to locate it up there. Thanks for the suggestion. I hadn't even thought of it.
posted by penduluum at 3:10 PM on November 30, 2008
I don't know if you already own the powerline kit, but you can actually buy powerline kits that include end points incorporating network switches (look for ones that have 4-port output on one end). Remember if you are buying that there are (currently) three standards - all marked "up to" (with a throughput of maybe 1/3-1/2 the "up to" value, if you're lucky) - 14MBps, 85MBps and 200MBps - so watch out for compatibility problems.
Even if you were to separate the XBox and server such that they couldn't easily share a single powerline output + switch combo, I don't think you'd need the layout in your second diagram. Instead, you could buy a second "output" adapter that pairs with the original powerline kit and gives you a third node.
CAT-5e/6 would be better than powerline, but may not be practical. I agree with anarchivist about not putting the server on an unnecessary powerline connection if it were avoidable.
posted by Nice Guy Mike at 4:38 PM on November 30, 2008
Even if you were to separate the XBox and server such that they couldn't easily share a single powerline output + switch combo, I don't think you'd need the layout in your second diagram. Instead, you could buy a second "output" adapter that pairs with the original powerline kit and gives you a third node.
CAT-5e/6 would be better than powerline, but may not be practical. I agree with anarchivist about not putting the server on an unnecessary powerline connection if it were avoidable.
posted by Nice Guy Mike at 4:38 PM on November 30, 2008
Response by poster: Yeah, unfortunately Cat6 is impractical; the two areas are too far apart, and there's no aesthetically pleasing way to run that much cable through my house. Too bad, too, because I have a pretty good sized chunk of Cat6 lying around in a box somewhere. Good to know about the adding-a-third-node thing, though. I didn't know if that was possible or not.
posted by penduluum at 4:54 PM on November 30, 2008
posted by penduluum at 4:54 PM on November 30, 2008
Best answer: I dont consider powerline a mature technology and for some people its just not the proper solution. It can sometimes have problems with bandwidth, packet loss, and high latencies. You need to do more than verifying the connection is up before you decide to go with it. You should run a ping for an hour from one device on one side of the powerline to another on the other side. Say File Server to Media server. Run this on the file server:
ping -t 192.168.1.100 (lets say this the IP for the media server)
In an hour or so press control-c to stop it.
You should only have 1 or 2% lost packets, if any. If you have over 10% lost packets then your connection is not stable and probably not good for gaming (or anything). It will also tell you the latency in ms. I could see a max of 20ms there. If you are regularly going over to 20ms then that might be a problem for gaming. If you are in the hundreds then the connection is not stable. If the pings drop out for seconds at a time the the connection will be worthless for gaming or large file transfers.
To determine your bandwidth you can use an application like iperf. You could run iperf -s on the media server and then iperf -c 192.168.1.100 on the file server. This will tell your usable bandwidth in mbps. Powerline ethernet real world speeds give around 4-15mbps of usable bandwidth. WirelessG about 3-12. Do several iperf tests as powerline fluctuates just as much as wireless, if not more.
Dont trust the ratings on the box, you need to do your own test. The speed rating on the box is the bus speed theoretical maximum. This does not include overhead, link quality, and interference. It is a legal/marketing/technical fiction. Both powerline and wireless have TONS of overhead. Regular ethernet has very little. That's why so many people run CAT5. Its just a lot more real world bandwidth. A 100mpbs wired connection is like 89 mbps of usable bandwidth.
So lets say youre getting a solid low latency 12mbps connection. That's pretty good for powerline. Thats tons of bandwidth for gaming and probably exceeds the speed of your internet connection.
>I eventually want to set this server up to stream media, serve to me through a VPN connection when I'm elsewhere, etc.
Well, that depends on your needs. If you need to stream HD quality materials then a 12mbps line will not cut it. You can do a slow copy of the file from one server to another, but you are not going to 'watch live.' SD yes. HD probably not. So you wont be able to stream a bluray rip but you can stream a dvd rip.
>What should that mystery box be? Would a switch work?
Yep a switch will be fine there.
posted by damn dirty ape at 5:26 PM on November 30, 2008
ping -t 192.168.1.100 (lets say this the IP for the media server)
In an hour or so press control-c to stop it.
You should only have 1 or 2% lost packets, if any. If you have over 10% lost packets then your connection is not stable and probably not good for gaming (or anything). It will also tell you the latency in ms. I could see a max of 20ms there. If you are regularly going over to 20ms then that might be a problem for gaming. If you are in the hundreds then the connection is not stable. If the pings drop out for seconds at a time the the connection will be worthless for gaming or large file transfers.
To determine your bandwidth you can use an application like iperf. You could run iperf -s on the media server and then iperf -c 192.168.1.100 on the file server. This will tell your usable bandwidth in mbps. Powerline ethernet real world speeds give around 4-15mbps of usable bandwidth. WirelessG about 3-12. Do several iperf tests as powerline fluctuates just as much as wireless, if not more.
Dont trust the ratings on the box, you need to do your own test. The speed rating on the box is the bus speed theoretical maximum. This does not include overhead, link quality, and interference. It is a legal/marketing/technical fiction. Both powerline and wireless have TONS of overhead. Regular ethernet has very little. That's why so many people run CAT5. Its just a lot more real world bandwidth. A 100mpbs wired connection is like 89 mbps of usable bandwidth.
So lets say youre getting a solid low latency 12mbps connection. That's pretty good for powerline. Thats tons of bandwidth for gaming and probably exceeds the speed of your internet connection.
>I eventually want to set this server up to stream media, serve to me through a VPN connection when I'm elsewhere, etc.
Well, that depends on your needs. If you need to stream HD quality materials then a 12mbps line will not cut it. You can do a slow copy of the file from one server to another, but you are not going to 'watch live.' SD yes. HD probably not. So you wont be able to stream a bluray rip but you can stream a dvd rip.
>What should that mystery box be? Would a switch work?
Yep a switch will be fine there.
posted by damn dirty ape at 5:26 PM on November 30, 2008
Actually my powerline ethernet info is a couple of years old. It looks like the newer models do much more than 12mbps and can sustain a 18mbps high def stream. Its still worth the time to do all the tests I mentioned earlier as every home is difference and just moving to a different plug can dramatically affect bandwidth.
posted by damn dirty ape at 5:41 PM on November 30, 2008
posted by damn dirty ape at 5:41 PM on November 30, 2008
I own a pair of netgear HD powerline modules. Don't believe the hype. latency is all over the place, and even under ideal conditions bandwidth is limited. I tested them with two of them on the same 4-plug strip and still couldn't steam 720p reliably. They're ok for basic connectivity - browsing and email access, slow file copying, that sort of thing, but HD streaming or gaming is not their forte, though it's certainly possible to do SD. Since you already own them though, it won't hurt to try.
Over the same sort of range, I've had much better performance with 802.11n wireless, especially 5Ghz - you could use a 802.11n WAP with a built in switch instead of the powerline + switch - but that does rely on being able to get a passable signal to the 360 + server.
If you do stick to powerline, it is possible to have more than one set on the same ringmain - you can either have three or more joined together as one network with multiple points as suggested, or four as two separate networks (you configure them with the app on CD), though they will both run slower as a result.
posted by ArkhanJG at 11:10 PM on November 30, 2008
Over the same sort of range, I've had much better performance with 802.11n wireless, especially 5Ghz - you could use a 802.11n WAP with a built in switch instead of the powerline + switch - but that does rely on being able to get a passable signal to the 360 + server.
If you do stick to powerline, it is possible to have more than one set on the same ringmain - you can either have three or more joined together as one network with multiple points as suggested, or four as two separate networks (you configure them with the app on CD), though they will both run slower as a result.
posted by ArkhanJG at 11:10 PM on November 30, 2008
If you did go wireless btw, the WAP would need to be able to be a client to your current wireless network. Most WAPs can also be a client (bridge mode), and are often cheaper and more flexible than dedicated wireless bridges.
posted by ArkhanJG at 11:14 PM on November 30, 2008
posted by ArkhanJG at 11:14 PM on November 30, 2008
Echoing what others have said: I have a pair of Netgear "HD" powerline adapters. Even plugged into the same circuit and less than 30 wire feet apart, they still exhibit problems if any other devices are on that circuit. For instance, I went from ~3% packet loss and a 21mbps (per the device's diagnostic tool) connection to ~88% loss and a 200kbps (again, per the tool) connection when I plugged in my TiVo to the same wall outlet.
Frankly, I thought it would be awesome, especially for my apartment where running wiring through walls is not possible. I ditched the idea and have now co-opted the (unused by me) telephone wiring (that's conveniently 4-pair Cat5 cabling) for my networking purposes.
Short story: Don't do powerline. If you can't do wired, use 802.11a wireless, as it runs on a different frequency that few others use.
posted by fireoyster at 7:14 AM on December 1, 2008
Frankly, I thought it would be awesome, especially for my apartment where running wiring through walls is not possible. I ditched the idea and have now co-opted the (unused by me) telephone wiring (that's conveniently 4-pair Cat5 cabling) for my networking purposes.
Short story: Don't do powerline. If you can't do wired, use 802.11a wireless, as it runs on a different frequency that few others use.
posted by fireoyster at 7:14 AM on December 1, 2008
Response by poster: Just in case anybody finds this thread looking for information on powerline networking: it definitely isn't for everybody.
But it works exceedingly well for me. In 20 iperf tests I got bandwidth results from 16.1 to 18.7 Mbits/sec, and over the course of a half hour I had latency results averaging 3 msec with zero lost packets. I understand it doesn't work well for many people. It works brilliantly in my house, much better than wireless (802.11g or n) does.
If anybody's curious, I think I'm going with a different setup than I outlined in my diagrams. I'm going to turn that "file server" into just a little barebones box that's hooked directly to the television, put something like XMBC on it and go with it. Put a bunch of hard-drives in it. I'll worry about being able to connect to it remotely sometime down the road.
posted by penduluum at 2:04 PM on December 4, 2008
But it works exceedingly well for me. In 20 iperf tests I got bandwidth results from 16.1 to 18.7 Mbits/sec, and over the course of a half hour I had latency results averaging 3 msec with zero lost packets. I understand it doesn't work well for many people. It works brilliantly in my house, much better than wireless (802.11g or n) does.
If anybody's curious, I think I'm going with a different setup than I outlined in my diagrams. I'm going to turn that "file server" into just a little barebones box that's hooked directly to the television, put something like XMBC on it and go with it. Put a bunch of hard-drives in it. I'll worry about being able to connect to it remotely sometime down the road.
posted by penduluum at 2:04 PM on December 4, 2008
This thread is closed to new comments.
posted by RustyBrooks at 2:32 PM on November 30, 2008