What should be the performance of a structured cabling system? Some
prioritize link performance, others quote channel performance. How can
installers and end users understand what performance they're really buying —
and be sure that that's what they get? First lets just recap what the
standards define by the words channel and link — and how they're not
necessarily what you might think.
The channel is the entire cabling system comprising all of the cable,
connecting hardware such as outlets and patch panels, and all the cords between
(but not including) the mated plug/socket in the network interface card (NIC),
used in your PC at one end — and the comms room equipment, typically a switch,
at the other. The performance requirement of the channel is actually handed down
to the cabling industry by the groups who define 'applications' such as
Ethernet, ATM, SDH and the like. These specifications basically say "If the
d.c. resistance is better than A Ohms, the insertion loss is better than B dB,
the signal to noise ratio is better than C dB, crosstalk is better than D dB,
the return loss is better than E dB, etc... then the application will work over
the channel". These groups that define the applications are only interested
that an RJ45 plug is at each end of the channel to fit into the RJ45 sockets
provided. They don't really care how the connection is made or configured as
long as the channel performance requirements are met.
Even
the different structured cabling standard bodies use different definitions! In
ISO/IEC11801:2002 and EN 50173-1:2002, a compliant channel can be any collection
of cables, connectors and cords that, as a cabling system, meet the channel
performance requirements. Whereas under the American standard, TIA/EIA-568-B ,a
channel must be constructed of individually compliant cabling components. And,
of course, if the individual components are not Cat 5e or Cat 6 component
compliant, then there is very little chance of the channel continuing to meet
its specified performance requirement should a different manufacturer's patch
cord, outlet jack or patch panel be introduced later on — leading to network
performance problems and reduced throughput.
A permanent link, as defined in ISO/IEC and EN, includes the horizontal
cabling from the horizontal patch panel jack in the communications room through
to the telecommunications outlet jack at the workplace — including the mated
connectors at each end. The permanent link is defined because, in most cases
when a cabling system is being installed, the channel (which also includes all
the additional cords and connections) simply doesn't exist at that point in
time.
Classes versus Categories
ISO/IEC 11801:2002 and EN 50173-1:2002 define various channels in terms of
cabling media and performance levels. For twisted pair cabling, there are three
high performance channels specified: Class D which is specified to 100 MHz
bandwidth and which is constructed using Cat 5e cabling components within what's
called a 'prescribed design'. A Class E channel is specified to 250 MHz, and
is constructed using Cat 6 cabling components. Similarly, a Class F channel is
specified to 600 MHz and is constructed using Cat 7 cabling components. The
possibility exists that channels can be constructed using proprietary cabling
components (cables, connecting hardware and cords), or from cabling components
that may look like Cat5e or Cat 6 components but are in fact not component
compliant. This can lead to poor (non- compliant) jacks and patchcords being
compensated for by better quality cable so that the overall channel meets or
exceeds the Class D, E or F performance requirements.
The
whole point of cabling standards is to be able to plug together cabling
components from different manufacturers and still meet the overall channel
requirement. You don't want a situation where an end user is mixing compliant
and non-compliant products. For instance, if manufacturer A makes its jacks with
a performance that is worse than the Cat 5e standard, but makes up for it in the
permanent link or channel by using a better performance cable, and manufacturer
B makes its jacks and cables so that they are only just compliant to Cat 5e (an
acceptable and very common practice), what happens when you mix manufacturer A's
jacks with manufacturer B's cable? Exactly — it's not going to work
properly! So, to ensure that cabling components from different manufacturers
will interoperate, we have Cat 5e and Cat 6 component requirements. Now, if all
the cable, outlet jacks, cords, and patch panels in a network are truly
component compliant, even if these cabling components are from mixed
manufacturers, not only do you have a system that will by definition give Class
D or Class E channel performance, in an ideal world you wouldn't even need to
test the system! Of course in the real world, it's quite possible for poor
terminations and wire map errors to occur or for jacks and plugs to become
damaged. And cables are all too easily crushed, kinked or damaged on site during
installation. So cabling components can cease to be component compliant with the
result that channel performance is jeopardised! In reality, it is imperative to
test on site after installation.
Of course there's a big problem here. Testing the channel or permanent link
does not reveal whether the components used are Cat 5e or Cat 6 compliant — it
can only tell you if, as installed, it meets the channel or permanent link
performance requirements. Do Cat 5e and Cat 6 cabling components guarantee Class
D and Class E performance respectively? Essentially, yes they do. It is the
intention of the standards that if you construct a permanent link using Cat5e or
Cat 6 cabling components it will meet the Class D or Class E requirements
respectively. If you then add fully component compliant Cat5e or Cat 6 cords and
connecting hardware respectively then you will have a Class D or E channel.
(And, in fact, if you mix Cat 5e and Cat 6 cabling components you will achieve a
minimum of Class D channel performance overall).
By definition, the electronics used with a Class D channel will never go past
100 MHz, and those using Class E will never go past 250 MHz. So manufacturers'
claims of 300/350/450 MHz are completely and totally irrelevant. It's just
pure, unnecessary, confusing hype. But let's be clear, although the standards
allow for a great diversity of proprietary solutions, 100% of customers want to
buy a system that is true Cat 5e or Cat 6 i.e. it's the fully interoperable,
fully component compliant option. And they can only achieve this by specifying
that all of the cabling components in the solution are fully component compliant
to the standards. If they are, and the permanent link and cords are fully
tested, the overall system will be exactly what they wanted to buy — a fully
compliant solution.
There
is a defining way to measure product capability and create a level playing field
and it's quite simple. Ensure that all of the cabling components in the system
offered compliant to Cat 5e (or Cat 6) according to ISO/IEC 11801:2002, EN
50173-1:2002 or TIA/EIA-568-B. That's what really matters. Users should decide
for themselves whether they value the comfort of a safety margin or headroom and
should totally discount irrelevant claims about additional bandwidth and channel
performance.
Bala K Chandran managing director Krone Communications India