The network is down…. again…. Some of the users are going ballistic. Others are telling cynical jokes about the company’s internal quality standards in the hallway. Although the network has grown to become critical to your company’s day-to-day business, the associated reliability standards have not
Instead of using yet another band-aid to fix the latest symptom, it is time to fix the root of the problem, but where do you start? What needs to be upgraded? In this thought process, cabling is often an afterthought, however, recent studies show that 40 per cent of all networking problems are attributable to cabling. And since structured cabling underpins today’s advanced networking infrastructure, bulletproof reliability can only be achieved with an end-to-end structured cabling design built for performance excellence that leaves nothing to chance.
Cabling Channel Performance Modeling
The underlying physics of many technical problems is very complex. Instead of dealing with all the complexity, models are often created that try to simplify the issues while maintaining the observed realities. Developing an accurate model is paramount to ensuring the conclusions drawn are applicable.
Recent work done by the Systimax R&D group at Avaya Labs has shown that the industry’s accepted cabling component model has major shortcomings in predicting channel performance. No two cabling channels in the field are exactly identical. They vary by cable length, cord length, number and type of connectors, connector direction (forward and reverse), etc. Hundreds of thousands of combinations are possible, and each combination can achieve different performance levels for different parameters. Some combinations of certain cabling components can result in dramatic and unexpected reduction in performance with an associated reduction in reliability.
Over the last decade, cabling standards have improved the requirements for electrical performance of individual components. Category 3 components were only specified for attenuation and NEXT. Category 6 components are specified for these as well as PSNEXT, return loss, FEXT, PSFEXT, delay and delay skew. All these parameters are important for balanced mode transmission. In fact, most of the balanced mode parameters for 4-pair cabling have now been specified. Standards committees have also developed models for predicting channel performance via cascading component performance. Some of these models are simple spreadsheet voltage summation models, while others are more complex random vector summation algorithms that take the statistical nature of cabling into account. However, all existing models to date have ignored cross-coupling and balance factors!
To fully characterize 4-pair cabling, the near-end and far-end coupling from each conductor to each other conductor must be characterized. A total of 256 parameters (16×16) must be accounted for. This is over 4 times the number of parameters currently specified in the relevant TIA, CENELEC, IEC, and ISO standards. The currently unspecified parameters can cause havoc in certain real world configurations of installed cabling channels.
Finding the worst case channel configuration before it finds you
The assumption is often made that a 4-connector 100-meter channel is the worst-case configuration for electrical performance, meaning if the 100-meter configuration passes, all other foreseeable configurations will also pass. This assumption is generally true for attenuation (more correctly referred to as insertion loss), FEXT, PSFEXT, and delay. It is not always true for return loss, and generally not true for NEXT and PSNEXT. The 100-meter assumption ignores the fact that NEXT can be significantly worse in short channel configurations. In extreme cases, it can even show up in configurations with only one connector on each end. The problem with short configurations was first observed in certain Category 5 links and referred to as “”Short Link Resonance””. After much investigation by the industry it was noted that energy exchange between two closely spaced connectors from certain vendors was the culprit. Products were improved to the point that the failure rate of Category 5 links decreased considerably. The problem went away for Category 5 only temporarily, and is starting to reappear in some Category 5E and many Category 6 links and channels.
Market migration to Category 6 and the resurgence of short link problems
Most companies who consider reliable networking as critical to their business already specify pre-standard Category 6 cabling. Others have taken a more conservative route and are waiting until the standard is ratified.
The Category 6 standard is nearing completion in both the TIA and ISO committees. In TIA, it is currently in the default ballot stage. This means that most of the document is “”frozen”” and no longer open for technical comments. There are still a few annexes open for comment, but these should be resolved in the next few months. The bottom line is that those waiting for ratification will not have to wait much longer, since 2002 will finally see the ratification of Category 6 standards worldwide.
Market research companies are generally in agreement that Category 6 will dominate the entire market (all enterprise wiring, independent of importance of reliability) in the next few years. The controversial issue is whether the dominant position will occur in one or three years. Gartner Group predicts 90per cent penetration rate in 2003. BISRIA does not predict 90 per cent penetration until 2005.
In part two of this two part series the article will get into the benefits of Category 6.