A regular 10Base-T or 100Base-TX connection only requires those two pairs, and yes, they are carried on pins 1/2 and 3/6 of the modular jack, usually connected using the white/orange and white/green pairs (see the recent 568A vs. 568B thread in this area for the debate over which way around).
The newer 1000Base-T networks require all four pairs, however, as do some 100Base-T variants which were intended to run over CAT3 cable. You're probably not likely to run into the latter though.
While this will work in a pinch it is NOT something that you want to install permanently. As with all IT geeks they often don't know much beyond the keyboard and putting this idea in a customer's head is plain wrong.
You can get adapters that plug into the 8P8C (RJ45 for the non-pedantic) jacks which have two 8P8C jacks on them with the appropriate pairs split out. Using two of these adapters at both ends will allow you to run two ethernet connections over one cable.
Or you could make your own, using three 8P8C jacks.
This avoids the need to modify the permanently-installed wiring.
Usage of these types of adapters is commonly done, when the cost to run another cable is prohibitive. I've seen it done at two different, very large, companies.
Same here. Standards are in place for a reason. Sure, you or your customer can split cable pairs out and "make it work". If you let the customer tell you how things are to be done, well you are already headed toward defeat. Do it right or don't do it at all is my motto.
Re-reading the original post, I'm getting the impression that this might be a planned new installation too.
My take is, I hope, the pragmatic one. If this is a completely new system, then pull two cables and have done with it. The cost of the cable compared to labor (both for initial install and any possible later wiring) is minimal. It's really no harder to pull two CAT5 cables than one.
However, if there is already a single link in place and a second is needed, if pulling another cable is going to involve considerable disruption and time, and if there's no chance that the existing link will need the extra pairs in the forseeable future, then I don't see a problem with "borrowing" the blue/brown pairs and simply transfering them to a second jack.
(RJ45 for the non-pedantic)
Well, I like to be a pedant. And I'm sure Ed will agree with me!
Thanks for the replies guys. As to new vs. existing install.... yes and yes. We originally pulled 12 drops in a second floor residential addition. A couple of the drops were specified as doubles, and that info was lost by the time it got to the installers. It was caught by the homeowner and I told him we would pull those additional cables. In the mean time he talked to his IT guy at the office who told him ethernet only uses 2 pair, so he decided don't worry about pulling additional cables and while we're at it, why not just make all the jacks doubles, just in case. So that's where we're at.
None of this is for anything specific, it is all just for future sake, as per owner. As far as I'm concerned there is no difference between pulling a jack out and plugging the port, or stealing a pair if needed, just that the modification is made in the reverse order as to what is needed, just that in doing this way is what the owner wants. Commercial, I would put up a lot more resistance to veering off the standard.
I would have used external adapters if only because they might want to upgrade to gigabit which uses all 4 pairs. But then again only a hardcore IT geek wants to run gigabit in their house
Truth be told, I mixed 100baseT and POTS telephone in the same cat5 cable for a residential installation I did for a friend of mine (surface wiring, none of it was fished). The IEEE 802.3 specs state that this is acceptable. (It is not acceptable for 100baseT4--was any 100baseT4 equipment actually ever sold?). It works fine. Running another cable was not an option due to the surface wiring.
Actually this is something I've always wanted to test anyway, since armchair electrical engineer wannabees always pontificate that the high ring voltage may interfere with the ethernet signals.
It works fine, but I wouldn't do it that way for anything but a residential installation with those contstraints, where I know that they'll never want to use gigabit ethernet over that cable, and where any potential failure is only going to cause their son's PC to not have internet access. (Not a big deal in the grand scheme of things).