Date: Wed, 21 Aug 2002 10:45:52 -0700 From: Kevin Stevens <Kevin_Stevens@pursued-with.net> To: "Jim McAtee" <jmcatee@mediaodyssey.com> Cc: "Work" <work@thuntek.net>, <freebsd-newbies@FreeBSD.ORG> Subject: Re: How important is Physical? Message-ID: <D674587E-B52D-11D6-8CFB-003065715DA8@pursued-with.net> In-Reply-To: <042b01c24937$18a3c490$272fa8ce@jim>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wednesday, August 21, 2002, at 10:20 , Jim McAtee wrote: >> Hello, 1st post. I work for an ISP that runs Freebsd unix and >> we also service several Office complexexes. From CATV background, >> we always measured the DB Signal strength to determine how far a cable >> can run, and how many units you can serve, without requireing >> amplification. In networking are these mattters important? If so, how >> many computers can you run out of 1 network, and what unit of measure >> do you use to determine signal strength? Also, how important is >> Grounding? In addition to the ground or neutral circuit that is the >> 3rd prong on a normal 110 volt plug? You can answer the in the group >> or to me at work@thuntek.net . thanks >> >> Mike Harris > > Are you just talking about ethernet networking? It's pretty much a > no-brainer. If you're looking at Cat5, just comply with the standards > of > less than 100 meters per run (including patch cables at either end) as > well > as some other wiring recommendations such as the minimum radius to cable > bends, and avoiding running cable near electrical wiring. Testing > equipment > is available to test and verify the compliance of individual runs. > There's > no physical limitation that I'm aware of (other than network bandwidth > and > switch speed) to the number of machines on a Cat5 ethernet network. The limit is 1024 on a single 10BT network, I believe. To OP: Yes, there are specs, and yes, things do break when you don't meet spec or install properly. However, most of the real-world measurable parameters has been codified in the specs - if you use CAT5 cable and stay within 100m you are pretty well guaranteed that things like near-end crosstalk and dB loss will be ok. However, they are still issues and you do need to be aware of them. Light loss budget issues are still calculated for fiber installs. Another thing to be aware of is that there are signalling requirements that are incorporated into those specs as well. For instance, on a plain 10B2 (coax) Ethernet segment, you may be able to extend physically well beyond the limit (185m if I recall) and get good signal. However, the network will fall apart because the collision-sensing mechanism won't work as the signal propagation is now too slow. IEEE 802.3 is the master standard for Ethernet, then there are various cabling standards such as TIA. A Google search will get you up to speed, and some of the manufacturers like Cisco also have cabling requirements papers available. A number of companies including Fluke have cable test equipment that is used to do the kind of monitoring you're describing. KeS To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-newbies" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?D674587E-B52D-11D6-8CFB-003065715DA8>