Connections +
Feature

Testing the 10GBASE-T Waters

Now that the IEEE has approved the standard, it is time to start thinking what this new generation of cabling will mean for the front line technicians


January 1, 2007  


Print this page

While there has been a lot of buzz around the eventual adoption of 10 Gigabit/sec Ethernet over twisted-pair copper cabling (10GBASE-T), it is only recently that it has been considered a viable technology for the real world.

It is still early in the game to see any real take-up of this new technology in Canada. Now that the IEEE (Institute of Electrical and Electronics Engineers) has approved a standard, it is time to start thinking ahead to what this new generation cabling will mean for the front line technicians.

Among other changes, 10GBASE-T will definitely present some new testing challenges. This includes a new phenomenon called Alien Crosstalk, which will require an added layer of testing expertise.

There is no question that over time, 10 Gigabit could become as commonplace as 1 Gigabit is today for data centre applications. In an effort to keep pace with growing demand for capacity, operations have already far exceeded the limits of 1 Gigabit backbones and are resorting to running multiple 1 Gigabit links and trunking them as a means to increase capacity.

This may serve current needs, but it is becoming increasingly complex and costly to manage this type of cabling infrastructure.

Given the escalating demand for bandwidth, a more widespread move to 10 Gigabit is inevitable.

Still a rare commodity

Before that happens, however, there are some issues that need to be addressed.

As it stands today, implementation of 10 Gb/s Ethernet over fiber is out of reach for most IT budgets. That means 10 Gigabit implementations are an extremely rare commodity in data centre operations today.

A large part of the expense lies in the interfaces for fiber. If we look back at the early evolution to 1 Gigabit, moving from fiber to copper will be instrumental in bringing the interface costs down, once we see widespread availability of copper interfaces.

The first step to the transition was the ratification of the 10GBASE-T standard. With this we can now expect to see some momentum building as enterprises start to examine if their infrastructures are ready for the increasingly bandwidth hungry applications of the future.

Many have come to realize that running multiple 1 Gigabit lines can only address this issue in the near-term. It will be 10GBASE-T that will prove easier and less costly in the long term, because it will dramatically reduce cabling requirements and the number of interfaces between servers and switches.

The transition to 10GBASE-T will not happen overnight. There are still some performance and testing challenges in running 10 Gigabit over Category 6 cabling to be resolved. A major challenge is distance limitations.

Cat 6 is only designed to work up to frequencies of 250 MHz. However, test frequency for 10 Gigabit has been extended to 500 MHz — and cabling working at a higher frequency is more susceptible to picking up noise from adjacent cables over longer distances. Beyond 37 metres, the likelihood of this type of noise — otherwise known as Alien Crosstalk — grows.

While all the familiar parameters for testing traditional Cat 6 will apply (now up to 500 MHz), Alien Crosstalk will become a new test requirement also requiring testing to 500 MHz.

Testing for Alien Crosstalk is an issue that has been examined by the IEEE and is currently being addressed by the Telecommunications Industry Association (TIA). The TIA will soon issue TSB155 (Telecommunications Systems Bulletin) outlining guidelines to field test Cat 6 for 10 Gigabit use.

Once these are ratified, the next step will be to establish design specifications and field-testing criteria for augmented Cat 6 (Cat 6A) that will ensure that a 10-Gigabit unshielded twisted pair cable will work out to 100 metres.

While it could take a year or more before Cat 6A standards are ratified, this allows technicians the time to understand the technology requirements and gain the upper hand on the appropriate testing procedures for this new medium.

Tackling Alien Crosstalk

Crosstalk measures signal coupling from one wire-pair to another within a twisted-pair cabling link. This kind of coupling can create an effect that is much like a noisy transmission line. In some cases, the noise induced by crosstalk will prevent a receiver from distinguishing the signal sent by the transmitter at the other end of the link.

In a twisted-pair cabling link, there are two effects that have an impact on signal quality. First, signals traveling from one end of the cable to the other degrade over distance. In other words, the longer the cable, the smaller the signal is at the receiving end.

At the same time, insertion loss (or attenuation) of the transmitted signal increases as the frequency of the signal increases. The combination of greater noise from adjacent cables with an attenuated signal could in fact make a 10 Gigabit network perform much slower and is a significant barrier to enabling full bandwidth availability for twisted pair cabling.

Near-End Crosstalk (NEXT) and Far-End Crosstalk (FEXT) measure the crosstalk signal that appears at the same end of the cabling link from which the test signal or disturbing signal is launched.

Testing for Alien Crosstalk follows a similar approach to that used for in-channel Near-End and Far-End Crosstalk, since it requires testing a signal generated at both the far and near end. Unlike NEXT and FEXT which are done on the same cable or wire pairs (in-channel), Alien Crosstalk is tested on adjacent cables.

Alien crosstalk is unique in that the crosstalk coupling occurs between wire-pairs in different, adjacent cabling links. It presents special challenges for testing twisted-pair cabling since there is a combined impact of many wire-pairs in the bundle upon the “victim” wire-pair being tested.

The standard specifies using a tester with 500 MHz bandwidth (minimum) and accuracy to Level IIIe or better.

A contractor can guarantee the tester is accurate if it is calibrated and tested to the standard’s accuracy requirements. Performance verification by independent testing laboratories such as UL Labs ensures the tester meets or exceeds these requirements.

Below are the basics of testing for Alien Crosstalk:

Setting it up: The setup for in-channel testing is the same as traditional Cat 6 testing, 100% of the cables require this test. However, the installer will need a special Alien Crosstalk kit and the use of a laptop at the test site.

Running the test: The in-channel tests are run (plot data enabled) and saved on the laptop, then all the computational work is done by the laptop connected to the analyzer while the ANEXT and AFEXT tests are run on the various disturber and victim links.

What to measure: In addition to the traditional in-channel tests to 500 MHz, all disturber cables will be tested against select disturbed (victim) links for ANEXT and AFEXT — also to 500 MHz.

Troubleshooting: It will be noticed immediately if the link fails. This allows the installer to correct the problem before continuing on each link and wire pair.

Testing the Limits: In most cases, testing Alien Crosstalk between all possible wire-pair combinations is simply not economically feasible. Practical field test results however have proven that Alien Crosstalk between cables in different bundles is non-existent or negligible.

Therefore, test methods should be applied to cabling within bundles or adjacent connections in close proximity on the patch panel. To properly test any one disturbed or “victim” link, all of the links that belong to the same bundle of the victim link must be included in the test procedure as disturber links.

Before looking at a practical approach to sampling method, it is important to familiarize yourself with both the recommendations in the10GBASE-T standard (IEEE), the proposed TSB155, and proposed Augmented Cat 6 (Cat 6A) cabling standard (EIA/TIA568B.2-10).

A practical strategy for Alien Crosstalk is foc
using on those installed links with the highest probability of failing.

If, for example, the longest links in an installation pass, all shorter links will very likely pass with better margins. It is also important to include all the links in the same bundles as the selected link, as well as links terminated in adjacent positions in the patch panel.

Despite the challenges of implement-ation and testing, 10 Gb/s Ethernet for twisted-pair cabling holds significant promise. While it will be a job that requires extensive working knowledge and training, it is the network technicians that will play a key role in ensuring that this new super fast networking technology works flawlessly.

Some may feel that the added complexity of testing for Alien Crosstalk can be avoided, but ultimately the best installations will be those that are properly and thoroughly tested. It is the only way to guarantee optimum performance and avoid problems over the long term.

Brad Masterson is product manager for Fluke Networks Canada and can be reached at brad.masterson@fluke.com