Feeder cables with 10 Gig?
Bookmark this page:
 Yahoo!
 Google
 Windows Live
 del.icio.us
 digg
 Netscape
|
|
||||||||||||||||||||||||||||||||||
|
Posted by BJ on July 25, 2006, 12:15 pm
Please log in for more thread options We're about to rewire our office with a high-grade, shielded Cat6 cable that is supposed to support 10gig up to 55 meters (yes, I know there's no guarantee since there's no ratified standard yet). Our hope is that this high-grade wire will eliminate the need to do another network rewire for at least another 10-15 years. But there's a good chance that our server room may move from it's current location in the next couple of years. We could always use feeder cables to extend all the cable runs from the old server room to the new, but my question is, should we? If we hope to someday run 10 gig over this copper, will a properly-done feeder cable be as good as a single unbroken run of the same cable over the same length? Do I run the risk of performance loss, increased exposure to EMI, or any other problems by using feeder cable in this scenario, or would I need to do another complete rewire to get the cable integrity I'd need for 10gig? I realize I'm trying to predict the future of a technology that barely even exists today, but I'm hoping someone with particular expertise or experience in this area can tell me, on a gut level, what they think. I'd like to make the choices now that stand the best chance of serving us well for the future. In case it matters, we'll be running PoE over this wire today, and I can't image that 10gig networks won't try to do the same thing. All comments appreciated, and thanks in advance. Bryan | ||||||||||||||||||||||||||||||||||
|
Posted by Walter Roberson on July 25, 2006, 1:29 pm
Please log in for more thread options I have no expertise in wiring, let alone wiring for very high speed, so the following might be wrong or irrelevant: What I have read is that often the greatest challenge for high bandwidth is in the connectors, as it is at the connectors that signal reflections are generated, and at which impedence matches become an issue, and at which phase distortions are most sharply introduced into signals. This would suggest that if your aim is cutting-edge bandwidth, that you would be better off with an unbroken run. I've read the odd tidbit about very high speed wire networks; one of the suggestions I've seen is that in the future, cable interconnects might no longer be passive electromechanical systems, and might instead require some amount of electronics (I don't recall if they would be passive or active), such as in order to reduce echoes. It wasn't a "repeater" as such, in that the signal would not be regenerated, but it was more than a simple capacitor. | ||||||||||||||||||||||||||||||||||
|
Posted by BJ on July 25, 2006, 5:52 pm
Please log in for more thread options
Great info, thanks; I may forward it to my cabling guys for their reactions, and post them back here. BJ >
>>If we hope to someday run 10 gig over this copper, will a properly-done
>>feeder cable be as good as a single unbroken run of the same cable over >>the >>same length? Do I run the risk of performance loss, increased exposure to >>EMI, or any other problems by using feeder cable in this scenario, or >>would >>I need to do another complete rewire to get the cable integrity I'd need >>for >>10gig? >
> I have no expertise in wiring, let alone wiring for very high speed, > so the following might be wrong or irrelevant: > > What I have read is that often the greatest challenge for high > bandwidth is in the connectors, as it is at the connectors that > signal reflections are generated, and at which impedence matches > become an issue, and at which phase distortions are most sharply > introduced into signals. > > This would suggest that if your aim is cutting-edge bandwidth, > that you would be better off with an unbroken run. > > I've read the odd tidbit about very high speed wire networks; one of > the suggestions I've seen is that in the future, cable interconnects > might no longer be passive electromechanical systems, and might instead > require some amount of electronics (I don't recall if they would > be passive or active), such as in order to reduce echoes. It wasn't > a "repeater" as such, in that the signal would not be regenerated, > but it was more than a simple capacitor. | ||||||||||||||||||||||||||||||||||
|
Posted by glen herrmannsfeldt on July 26, 2006, 1:28 am
Please log in for more thread options
Walter Roberson wrote: >>If we hope to someday run 10 gig over this copper, will a properly-done
>>feeder cable be as good as a single unbroken run of the same cable over the >>same length? Do I run the risk of performance loss, increased exposure to >>EMI, or any other problems by using feeder cable in this scenario, or would >>I need to do another complete rewire to get the cable integrity I'd need for >>10gig? > I have no expertise in wiring, let alone wiring for very high speed,
> so the following might be wrong or irrelevant: > What I have read is that often the greatest challenge for high
> bandwidth is in the connectors, as it is at the connectors that > signal reflections are generated, and at which impedence matches > become an issue, and at which phase distortions are most sharply > introduced into signals. Well most people (except ethernet engineers) use coaxial cable when they need high bandwidth. It isn't so hard to make an impedance matched coax connector if the connector parts have the appropriate diameter. It is somewhat harder for UTP. > This would suggest that if your aim is cutting-edge bandwidth,
> that you would be better off with an unbroken run. I agree, though it might be that with good connectors the reduction in length is not so big. > I've read the odd tidbit about very high speed wire networks; one of
> the suggestions I've seen is that in the future, cable interconnects > might no longer be passive electromechanical systems, and might instead > require some amount of electronics (I don't recall if they would > be passive or active), such as in order to reduce echoes. It wasn't > a "repeater" as such, in that the signal would not be regenerated, > but it was more than a simple capacitor. Active terminators are popular for SCSI termination. I suppose that could soon be true for network cabling. -- glen | ||||||||||||||||||||||||||||||||||
|
Posted by William P.N. Smith on July 25, 2006, 1:47 pm
Please log in for more thread options
>We're about to rewire our office with a high-grade, shielded Cat6 cable
Why shielded? >If we hope to someday run 10 gig over this copper, will a properly-done
>feeder cable be as good as a single unbroken run of the same cable over the >same length? It can't be as good, as there'll be another couple of sets of connectors and a patch cable between them (or at least a Cat6 splice). >In case it matters, we'll be running PoE over this
>wire today, and I can't image that 10gig networks won't try to do the same >thing. An unknown spec for 10G, unknown PoE interactions, adding splices, "10G PoE midspans" (whatever they might look like), I think you are planning too far into the unforseeable future. I can't even buy 1G PoE midspans today... | ||||||||||||||||||||||||||||||||||
| Similar Threads | Posted |
| Feeder cables with 10 Gig? | July 25, 2006, 12:15 pm |
| Supply Cat5e Patch Cords,Cat5e Patch Cables,Network Patch Cables,Network Patch Cords,Unshielded Patch Cables,Cat6 Patch Cables,Cat6 Patch Cords | April 12, 2006, 12:42 am |
| STP cables | October 9, 2004, 8:41 am |
| Patch cables for 1000BaseT Cat? | October 26, 2004, 2:48 pm |
| Crossover Vs. Patch Cables | March 9, 2006, 11:15 am |
| making cables - what am I doing wrong? | April 2, 2006, 10:21 pm |
| most common twisted pair cables | February 23, 2007, 8:53 am |
| 802.3ad Link Aggregation between two hosts using crossover cables | January 24, 2005, 11:23 am |
| Newbie: how to organize cables on a patch panel? | June 3, 2007, 6:28 pm |
>feeder cable be as good as a single unbroken run of the same cable over the
>same length? Do I run the risk of performance loss, increased exposure to
>EMI, or any other problems by using feeder cable in this scenario, or would
>I need to do another complete rewire to get the cable integrity I'd need for
>10gig?