After conducting some research for the article discussing Ethernet and InfiniBand that appears on page 1, I Tweeted about one of today’s most intriguing Ethernet topics–10GBase-T. The twisted-pair flavor of 10-Gigabit networking looks poised to overcome some hurdles that have slowed its path to adoption, in sharp contrast to the speed with which the protocol will move 1’s and 0’s.
In a way that only Twitter can force me to, I summarized the situation as follows: “Proponents of 10GBase-T say barriers to adoption–price and power consumption–are coming down.” The post generated a few replies. Among them were, “Until the power consumption gets very close to fiber 10GBase-T will remain an odd niche product. No room for power hogs,” and, “But I see many customers putting fiber in the rack in anticipation of 40/100G.” To that comment came, “Correct, why not then 10GBaseT from server LOM to top of rack switch? 100/1G/10G server LOM big reason 10GT more than niche.”
It was good to see a healthy discussion on the topic, and satisfying for me to see that at least someone sees a future for 10GBase-T as more than just “an odd niche product.” Having followed the structured cabling industry for a number of years, I now see the RJ-45 (or, more accurately, the 8-position 8-contact/8P8C) interface in my sleep. Any user who also sees the RJ-45 in everyday use knows the performance level of its twisted-pair cable and connectivity system. As a result, the user has some idea of the networking protocol(s) the cabling system will support--potentially up to and including 10-Gbit Ethernet.
10GBase-T may be considered a crowning achievement for designers of the RJ-45, as well as the electronics that send signals through the interface that is practically ubiquitous in data networking. Not too many years ago, some were saying the twisted-pair version of 10-GbE would never happen. It’s happening now. Just in time to hear that it will never, ever handle 40- or 100-Gbit/sec transmission.
And perhaps it won’t. The hurdles to 10GBase-T adoption are coming down; they’re not down yet. We are still several months away from lower-power-consumption 10GBase-T devices hitting the market. And as this issue’s article points out, InfiniBand has spent the past decade making a play for short-run, high-speed connectivity in environments such as data centers. The RJ-45 has done everything it’s been asked to do so far. Will it do more? When I recently posed that question to a manufacturing company that produces about a million connection points daily–many but not all of them RJ-45s–they had a Twitter-ish response: Never say never.
I’ve said this before but I’ll say it again because it’s still true: Travel overseas and you’d better either bring or buy various adapters in order to plug your electronic devices into electrical outlets. But that RJ-45 patch cord in your travel bag will connect you to the network no matter where you are. Someday soon, with the advent of Power over Ethernet Plus, that RJ-45 may also carry the direct-current electrical power your laptop will need. Despite the inherent limitation of a built-in split pair, the RJ-45 has exceeded the performance expectations of some by meeting the performance requirements of so many users.
Enough of what I think. What do you think? If you can put it in 140 or fewer characters, Tweet me: twitter.com/cablingmag
 |
Patrick McLaughlin
Chief Editor
Patrick@pennwell.com
