COPPER CABLING
Given that it is far easier to design the electronics and cabling system for a system capable of 70m or 50m, and that it is possible to design a data centre with all links 70m or less (and use fibre for any really long ones) this poses the question of whether 100m is a ‘sensible’ requirement this time around. “But we’ve always done it this way!” is a very bad premise from which to start any new project! A pattern has previously emerged, which looks set to repeat itself with 40/100Gb/s. It’s this: Fibre gets there first with the new higher speed. It’s very expensive at first – but the cost is justifiable for those leading edge applications that need the new higher speed long before it becomes mainstream. And for 40 and 100Gb/s, we’ve already seen the fibre standards published and systems becoming available.
With faster fibres comes short-reach copper for equipment backplanes and short interconnects. That’s because the electronics needed for copper transmission on short-reach is simpler than for long channels and more expensive high-performance cable and connectors can be justified. With short reach Gigabit Ethernet, the limit was 25m on twinax cable. By 10Gb/s, this was down to 15m and the recently defined standards for 40Gb/s and 100Gb/s it is 7m maximum. But it’s good for early requirements. Historically, a couple of years later, along comes the new generation of BASE-T. The cost of the copper channel soon becomes a mere fraction of that for both fibre and short-reach copper. At which point, the short-reach copper becomes obsolescent and the fibre reverts to being a solution for longer distances. I predict that we will see exactly this same process with 40Gb/s and 100Gb/s – although they may both go through the process in parallel.
DM: Is UTP dead?
KH: You might think, as the 10Gb/s market has now become skewed towards shielded and the issues of alien crosstalk are likely to be orders of magnitude worse at 40/100Gb/s than at 10Gb/s, that UTP would be out of the question. And many in the industry are saying so. However, I’m definitely not counting it, or maybe a hybrid solution out of the running. I’ve seen recent papers from highly reputable engineers indicating that UTP could be possible even at these esoteric speeds. And for those who remember the early development of 10Gb/s, it was said by many to be impossible then too.
DM: How will 40G/100G work? No-one can know how 40G/100G will work yet. 26 NETCOMMS europe Volume III Issue 1 2012
KH: With the exception of the 40Gb/s and 100Gb/s short reach copper solutions, it is essential to understand that despite the marketing claims of some manufacturers, no-one can possibly know! We in the R&D world all know lots of ways they could work. But these use many different combinations of cable
characteristics drawn from cable types that already exist and those which are not yet even designed, together with multiple options for line-coding and digital signal processing (DSP). There are dozens, if not hundreds of possible permutations. No-one can yet predict which will win out technically, commercially and politically. In fact, the IEEE working group from whom the technology solution will be developed (and this usually takes more than two years in the development phase alone) has only just been formed. The solution is not on the table yet. So, if any vendor is currently trying to sell you a ‘40Gb/s or 100Gb/s ready’ copper cabling system, be very, very cautious.
DM: Can existing structured cabling support 40G?
KH: The only honest answer is that no-one knows. Cat6A seems an unlikely choice, but the RJ45 footprint is preferred and silicon designers may be able to use the Cat6A performance envelope over shorter distances. Cat7 is unlikely to be chosen. Cat7A has a slim possibility of being the preferred media, but it’s more likely to be a revised Cat7A (more of a Cat7AA perhaps) operating at higher frequencies instead of the current 1000MHz, 1600MHz or even higher frequency may be required. Our own modelling based on Technology Forecasting Techniques developed with DeMontfort University indicates that a 2000MHz cable frequency could actually prove to be needed. It’s going to be very interesting while all of these technology options are considered by the IEEE study.
DM: Which connector will it use?
KH: So far, the industry has managed to re-invent the RJ-45 connector at every stage of Ethernet’s evolution. From 10Mb/s to 100 Mb/s to 1Gb/s and currently 10Gb/s, the RJ-45 has remained the connector of choice. Other connectors have been designed along the way with potentially higher bandwidth capabilities, with one exception; the ISO/IEC 60603-7-71, which combines conventional RJ-45 for up to Gigabit with ‘four corner’ pairs
www.netcommseurope.com
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48