Sigh...
Another article poorly researched and/or no proof-reading from Tech Republic.
1. Shielded Twisted Pair is not necessarily plenum-rated. STP is typically used for token-ring without baluns or SLIP for dumb terminals. Plenum cable is teflon-jacketed so that it doesn't generate toxic smoke in a fire. Some cities, notably Chicago specifically disallow plenum cable and require data wiring to be enclosed in conduit so the smoke is contained in the conduit. The plenum-rated cable isonly needed where it is used in air-handling spaces such as a ceiling without air-return ducts. Note that teflon jackets and foil shields inside the jackets are not the same.
2. When daisy-chaining hubs and switches use the 5-4-3 rule within a single collision domain. More importantly, look at how much traffic is being condensed into the daisy-chain port to ensure you don't risk collision traffic.
3. Cheap NIC's aren't what generate collisions. Poorly cascaded hubs and switches are the primary culprit, unused protocols are next in the lineup. If you have migrated from Novell to NT, none of your clients should have IPX loaded, and certainly nobody should have NetBEUI. This is basically poor physical network management.
To properly ascertain the likelihood of too much collision traffic you must measure your existing traffic and allow for future growth within the timeframe you expect your current installation to live. There is no magic calculation here, you have to know your network and its growth potential and plans. Most books advocate no more than 10% collisions. My networks generally run 1 collision per million frames or less because I'm fully switched.
Let's try to keep the articles at a more professional level. Spreading incorrect information helps nobody.
Discussion on:
View:
Show:
First it should be noted that the 5-4-3 rule only applies to 10Mbps Ethernet not 100Mbps.
The 5-4-3 rule was originally designed for ethernet bus networks. These networks use either thinnet (RG-58 co-axial) or thicknet (RG-8 co-axial, I think), and correspond with Ethernet standards 10base2 and 10base5 respectively.
The rule states that there can be 5 segments connected by 4 repeaters, and 3 of the five segments can be populated (have computers on them). The two empty segments can be usedextend the network to a distant location.
The whole rule is meant to ensure proper timing on a network. If it is two far from two end points on the network then when computers do a carrier sense there is the potential that a computer at the far end of the network has started talking but the computer at this end has not recieved the signal yet. Overall the result is excess collisions start to happen on the network, and the network slows down as it recovers from the collisions.
With 10baseTEthernet this rule effectively works out to a system where you can daisy chain 4 hubs and stay within the bounds of the 5-4-3 rule.
With 100baseTX you are limited by the class of hub that you buy. Class I hubs are not supposed to be daisy chainedat all, while class II hubs can have two hubs daisy chained together. These are much more restricted due to the more precise timing required by 100Mbps.
I assume that daisy chaining is simply not allowed with 1Gbps.
You would also be amazed how many people are not aware of the 5-4-3 rule.
The 5-4-3 rule was originally designed for ethernet bus networks. These networks use either thinnet (RG-58 co-axial) or thicknet (RG-8 co-axial, I think), and correspond with Ethernet standards 10base2 and 10base5 respectively.
The rule states that there can be 5 segments connected by 4 repeaters, and 3 of the five segments can be populated (have computers on them). The two empty segments can be usedextend the network to a distant location.
The whole rule is meant to ensure proper timing on a network. If it is two far from two end points on the network then when computers do a carrier sense there is the potential that a computer at the far end of the network has started talking but the computer at this end has not recieved the signal yet. Overall the result is excess collisions start to happen on the network, and the network slows down as it recovers from the collisions.
With 10baseTEthernet this rule effectively works out to a system where you can daisy chain 4 hubs and stay within the bounds of the 5-4-3 rule.
With 100baseTX you are limited by the class of hub that you buy. Class I hubs are not supposed to be daisy chainedat all, while class II hubs can have two hubs daisy chained together. These are much more restricted due to the more precise timing required by 100Mbps.
I assume that daisy chaining is simply not allowed with 1Gbps.
You would also be amazed how many people are not aware of the 5-4-3 rule.
The wiring setup TechRepublic had was also wrong for both 568A and 568B, the 568A should be brown, brown-stripe, orange, orange-stripe, blue, blue-stripe, and green, green-stripe.
The 568B should be brown, brown-stripe, green, green-stripe, blue, blue-stripe, and orange, orange-stripe. C'mon people!
The 568B should be brown, brown-stripe, green, green-stripe, blue, blue-stripe, and orange, orange-stripe. C'mon people!
I was thinking of a connector and not the cat5 cable itself. They were right, my bad. 
It's this type of misinformation that causes chain-reaction problems in our industry. Knowing when to install plenum-rated twisted-pair is dependant on the air return system in a building and is directly related to local fire codes. It has nothing to do with UTP/STP. You can obtain either with plenum-rated jackets. Somebody has to proof-read these articles!
- Keyboard Shortcuts:
- Prev
- Next
- Toggle
