#361369 - 14/04/2014 18:37
Best Practices: Long ethernet runs
|
carpal tunnel
Registered: 19/01/2002
Posts: 3584
Loc: Columbus, OH
|
For those of you with business networking experience -
I'm in the middle of overhauling a small campus network and putting in a new wireless network.
I have about a 250m run to an AP in another building. I've currently have two cheap switches in between that do nothing but boost the power and move the packets along. Of course, between the two switches, the two UPS's they are plugged into, and the AP, I have five points of possible failure that could take the network down.
Are there any best practices to eliminate some of these possible failure points, or any specific equipment you recommend? The AP at the end is a Ubiquiti UniFi which uses a non-standard POE spec, so it has a midstream power injector which is yet another possible point of failure. Is there any standard convention to differentiate regular ethernet cables from ones with POE, like a particular color cable?
EDIT: I should mention that fiber optic isn't an option unless it can be done fairly reasonably by yours truly. I'm doing this on a smallish budget as a volunteer for a church.
_________________________
~ John
|
Top
|
|
|
|
#361370 - 14/04/2014 22:06
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
old hand
Registered: 29/05/2002
Posts: 799
Loc: near Toronto, Ontario, Canada
|
What is the wireline speed?
10mbit or 100 or Gigabit Ethernet?
Is the cable already in place? Which CAT spec?
250 meters line of sight or 250 meters cable length?
Edited by K447 (14/04/2014 22:07)
|
Top
|
|
|
|
#361371 - 14/04/2014 22:59
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
pooh-bah
Registered: 20/01/2002
Posts: 2085
Loc: New Orleans, LA
|
Cheaper 100M POE uses the non-used pins for power, or type b. Gigabit uses all 8 for data so it has to use the more complicated type a, Common-mode signal, which puts voltage across all 8 pins using a common ground.
|
Top
|
|
|
|
#361374 - 15/04/2014 01:13
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
carpal tunnel
Registered: 19/01/2002
Posts: 3584
Loc: Columbus, OH
|
The existing cable is Cat5 or Cat5e. I'll check tomorrow.
100 mbps network.
Roughly 250m cable length, but perhaps less. It's less than 200m directly.
_________________________
~ John
|
Top
|
|
|
|
#361376 - 15/04/2014 12:21
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
old hand
Registered: 27/02/2003
Posts: 777
Loc: Washington, DC metro
|
|
Top
|
|
|
|
#361378 - 15/04/2014 17:09
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
carpal tunnel
Registered: 08/03/2000
Posts: 12344
Loc: Sterling, VA
|
John,
My apologies if I didn't mention the non-standard POE on the Unifi when you asked me about the product. The non-standard part of it doesn't have anything to do with the ethernet cables, though, just with the source of the power. I believe the Unifi products use 24V instead of the standard 48V (but I'm not very read up on POE stuff). Ubiquiti sells POE switches, so if you wanted to replace the last switch in your setup you could do that.
For such a long run, the alternative would probably be fiber, but then you'd have fiber adapters at either end, leaving you with about the same amount of equipment as you had before. I'd say if your current configuration is working for you, then leave it alone.
_________________________
Matt
|
Top
|
|
|
|
#361380 - 15/04/2014 18:40
Re: Best Practices: Long ethernet runs
[Re: jmwking]
|
old hand
Registered: 29/05/2002
Posts: 799
Loc: near Toronto, Ontario, Canada
|
Dual-Comm 10/100Base-T Ethernet over Coax Adapter Kit
Model Number: DECA-100 ... Passive device. Small size and light weight How can it be purely passive at both ends? There are several variations of ethernet over coax. Say, do you need an actual WiFi Access point at the far end, or just a WiFi antenna?
Edited by K447 (15/04/2014 18:47)
|
Top
|
|
|
|
#361381 - 15/04/2014 20:10
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
old hand
Registered: 27/02/2003
Posts: 777
Loc: Washington, DC metro
|
That is such a cool idea. I'm trying to figure the ideal implementation - a building with a lot of stone, concrete, and steel - and also pre-wired for cable. Small n.
Years ago, I had to temporarily extend a network between floors in an old building south of London. Stone walls and foot-thick concrete floors, with no useful telecom shafts between. The landlord didn't want us drilling holes ("Really, your next tenant will love it!") or using the stairwell. We finally just dropped a cable out one window and into the next. Ugly, but functional. <sigh> Gotta do what you gotta do.
As for the passivity - I suspect it's in that coax is far better at holding a signal, and can induce enough followup signal at the other end. But I'm not an EE.
-jk
|
Top
|
|
|
|
#361395 - 18/04/2014 01:33
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
carpal tunnel
Registered: 19/01/2002
Posts: 3584
Loc: Columbus, OH
|
Thanks for the suggestions guys. That coax adapter looks attractive and the money is right. Also nice that I don't have to mess with power and UPS's.
Any standard convention to distinguish cables that are POE so that people don't plug them into non-POE devices?
_________________________
~ John
|
Top
|
|
|
|
#361398 - 18/04/2014 12:54
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
old hand
Registered: 29/05/2002
Posts: 799
Loc: near Toronto, Ontario, Canada
|
Thanks for the suggestions guys. That coax adapter looks attractive and the money is right. Also nice that I don't have to mess with power and UPS's.
.... Did you find any decent reviews of the particular coax gear that validate the manufacturer performance claims?
|
Top
|
|
|
|
#361400 - 18/04/2014 13:35
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14496
Loc: Canada
|
The thing with long coax runs is signal loss. With modern RG6 75-ohm TV coax, half the signal power is lost for every 50' of cable.
Dunno what the specs are on whatever coax is being considered here, but it might not work out as planned, especially for two-way WIFI. It's often easy to boost transmit power to compensate for long coax runs, but that doesn't help with the other direction: reception.
|
Top
|
|
|
|
#361402 - 18/04/2014 13:59
Re: Best Practices: Long ethernet runs
[Re: mlord]
|
carpal tunnel
Registered: 08/03/2000
Posts: 12344
Loc: Sterling, VA
|
The thing with long coax runs is signal loss. With modern RG6 75-ohm TV coax, half the signal power is lost for every 50' of cable.
Dunno what the specs are on whatever coax is being considered here, but it might not work out as planned, especially for two-way WIFI. It's often easy to boost transmit power to compensate for long coax runs, but that doesn't help with the other direction: reception. I'd worry about that too. And don't those products assume that you have a single coax cable between both ends? I would assume that any lines in a building are going back to a splitter. Are you running your own coax, John? I've never terminated fiber myself, but I've seen it done. It's a bit of a nerve wracking process because the strands are so delicate. Plus you need powered adapters at each end, but they're very reliable in my experience. None of that matters, I guess, because I can't seem to find any fiber termination kits that aren't terribly expensive... As far as I understand it, you don't have to worry about someone plugging into a POE port. They're not going to fry their laptop. Someone else can correct me if I'm wrong... Personally, I don't like the coax idea. I think it's not likely to give you very good performance given that it's so passive. Maybe if you had Moca adapters on the ends supplying the power...
_________________________
Matt
|
Top
|
|
|
|
#361404 - 18/04/2014 23:50
Re: Best Practices: Long ethernet runs
[Re: Dignan]
|
old hand
Registered: 29/05/2002
Posts: 799
Loc: near Toronto, Ontario, Canada
|
The thing with long coax runs is signal loss. With modern RG6 75-ohm TV coax, half the signal power is lost for every 50' of cable... ... Maybe if you had Moca adapters on the ends supplying the power... Since there will be a (presumably) powered WiFi Access Point at the far end when powered MoCa adapters are used, that was my original line of thought, not the passive coax gear. Signal loss is not much an issue (within reason) when there are powered coax adapters at each end, especially if the coax is not split or otherwise running numerous places other than the Ethernet destination ends.
|
Top
|
|
|
|
#361405 - 19/04/2014 09:45
Re: Best Practices: Long ethernet runs
[Re: K447]
|
carpal tunnel
Registered: 08/03/2000
Posts: 12344
Loc: Sterling, VA
|
The thing with long coax runs is signal loss. With modern RG6 75-ohm TV coax, half the signal power is lost for every 50' of cable... ... Maybe if you had Moca adapters on the ends supplying the power... Since there will be a (presumably) powered WiFi Access Point at the far end when powered MoCa adapters are used, that was my original line of thought, not the passive coax gear. Ah ok. I didn't see that in your posts, and I was curious that the Wikipedia article you linked to doesn't even mention moca.
_________________________
Matt
|
Top
|
|
|
|
#361406 - 19/04/2014 10:20
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
carpal tunnel
Registered: 19/01/2002
Posts: 3584
Loc: Columbus, OH
|
If I decide to run the coax, it will be a single point to point run that I will do and terminate myself. It will be a simple matter of taping it off to the existing ethernet run and pulling it through the conduit between the buildings.
_________________________
~ John
|
Top
|
|
|
|
#361407 - 19/04/2014 10:46
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
carpal tunnel
Registered: 08/03/2000
Posts: 12344
Loc: Sterling, VA
|
Hmm, I figured it would be a good idea to look up maximum distances for MoCa, and it appears that it's similar to ethernet...So are there multiple conduits between the buildings? I'm wondering how you were able to get two switches in the middle of this run...
_________________________
Matt
|
Top
|
|
|
|
#361408 - 19/04/2014 15:36
Re: Best Practices: Long ethernet runs
[Re: Dignan]
|
old hand
Registered: 29/05/2002
Posts: 799
Loc: near Toronto, Ontario, Canada
|
Hmm, I figured it would be a good idea to look up maximum distances for MoCa, and it appears that it's similar to ethernet...So are there multiple conduits between the buildings? I'm wondering how you were able to get two switches in the middle of this run... I also forgot to check that. Probably has something to do with maximum signal round trip time within the ethernet low level protocol. There are Ethernet extender products that work beyond the 100 meter nominal ethernet maximum but I do not have experience with them. Some of this tech seems deprecated, while EPON Protocol over Coax seems to be a work in progress.
Edited by K447 (19/04/2014 15:41)
|
Top
|
|
|
|
#361417 - 20/04/2014 01:59
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
carpal tunnel
Registered: 08/03/2000
Posts: 12344
Loc: Sterling, VA
|
John, I have two questions:
1) is there direct line of sight between the two buildings?
2) have you thought about wireless?
Ubiquiti might have some cheap products to make a wireless uplink. I've never done it and I'm not sure I'm looking at their products correctly...
_________________________
Matt
|
Top
|
|
|
|
#361419 - 20/04/2014 11:02
Re: Best Practices: Long ethernet runs
[Re: JBjorgen]
|
carpal tunnel
Registered: 08/03/2000
Posts: 12344
Loc: Sterling, VA
|
To follow up, I think this might be what I'm talking about. You could use two of those to create the connection, or if the building at the other end is small enough and the signal reaches everywhere you need it, that dish might be all you need for an access point. I've never worked with these devices before, but they can easily cover the distance you need (they're rated for 30km, though I doubt that's real-world), and they're cheap. You just need line of sight for the two endpoints.
_________________________
Matt
|
Top
|
|
|
|
|
|