[qi:027] Copper, some telecom insiders joke, is like a cockroach – it never dies and keeps coming back. Someone always comes up with a way to use the copper in a new sort of a way. When we are all dreaming of a fiber to the home future, there comes an Australian academic who says that it is possible to send 250 Gigabits megabits per second over copper-based telephone lines.
And if that sounded incredulous, now engineers at Penn State University are claiming that it is possible to send 100 Gigabits per second (Gbps) over a distance of 100 meters using Category-7 (CAT7) cables that have four pairs of twisted copper wires inside. CAT7 cables were created to send data at speeds of 10 Gbps over 100 meters.
Penn State researchers presented this new technology in from of IEEE High Speed Study Group in Atlanta yesterday. The researchers developed transmitters and receivers that tried to overcome technical challenges such as crosstalk and interference.
These Category-7 copper cables used in the experiment are made by Nexans, a Paris-based company. They are being touted as the next generation Ethernet cables. More common Ethernet cables are called Cat5 cables. Penn State researches believe that sometime in the near future, it would be possible to build very high-speed local area networks and parallel computers.
When will this technology come to market? Lets just say by the time I would have lost all my remaining hair.
The researchers believe that two or three generations in the future, the technology of chip circuitry will allow these modem designs to be built. Currently, chip design is at about 65 nanometers, but they expect in the next two generations to get to what is required
The Australian guys hit 250 Mbps, not Gbps, although that would be cool!!
Yes as DEC mentioned above, you may want to correct your typo… it is not 250 Gps but 250 Mps instead (in that link to the Austrlian article).
I have to also state that making cables capable of that speed is one thing but it all depends on the interfaces at both ends. If you didn’t know, Cat6 copper cables out there currently support up to 1000 Mbps but no commercial NIC out there on the market currently supports it i.e. you can just go out to a Best Buy to buy one. So what I am trying to say is the speed the data sent down the cable is only as good as the interfaces on the ends (no matter what the capability of the cable is).
Unless the major manufacturers buy into the fact that it is profitable to commercially mass produce interfaces that support so called 100 Gps Cat7 speeds, we’ll probably never see it. My opinion is that if we haven’t seen 1000 Mbps NICs sold commonly now for the average user, we probably will never see 100 Gps ones. Most manufacturers probably see the future in fiber instead of copper and are reluctant to make the initial investment.
Yes as DEC mentioned above, you may want to correct your typo… it is not 250 Gps but 250 Mps instead (in that link to the Austrlian article).
I have to also state that making cables capable of that speed is one thing but it all depends on the interfaces at both ends. If you didn’t know, Cat6 copper cables out there currently support up to 1000 Mbps but no commercial NIC out there on the market currently supports it i.e. you can just go out to a Best Buy to buy one. So what I am trying to say is the speed the data sent down the cable is only as good as the interfaces on the ends (no matter what the capability of the cable is).
Unless the major manufacturers buy into the fact that it is profitable to commercially mass produce interfaces that support so called 100 Gps Cat7 speeds, we’ll probably never see it. My opinion is that if we haven’t seen 1000 Mbps NICs sold commonly now for the average user, we probably will never see 100 Gps ones. Most manufacturers probably see the future in fiber instead of copper and are reluctant to make the initial investment.
Om, most of what’s going in the field today is Cat5e. There’s very little Cat6 and when I do see it it’s only on brand new installs. The other problem with Cat6 is the requirements for maintaining the twist aren’t always understood by the installers. If the cablers don’t know what they’re doing it’s easy to get out of spec on Cat6. I can only imagine the challenges with cabling Cat7.
Also, you mention that Cat7 has 4 pairs of twisted wires. That’s true of all of these. The big difference is that each pair is shielded and then the bundle is shielded as well. Shielding adds $$$. Add in the cost of a different connector architecture and it’s easy to understand why it will take several years before anyone gives this serious consideration.
Still, copper is the cockroach because it’s magnitudes easier to install than fiber. Fiber requires very specialized (ie, expensive) tools to terminate and test. Copper’s also more resilient. You can really toss it around without worrying much about getting a kink in it. Try that with fiber!
And finally, it’s a lot harder to send power over fiber. 🙂
Foo, is there a typo in there? 1Gbps interfaces are pretty common these days. Only cheap PC’s still have 10/100 only interfaces.
Now, you’ll get no argument from me that the OS’s aren’t capable of supporting that kind of throughput.
However, with speeds like this it always starts in the backbone between core network devices before it trickles down to the end user. Some big iron like a Cisco 6500 might be able to push 100gbps in a few years. They are already pushing x10gbps without trouble.
I was actually inferring to the average computer, in the current trend of cheap PCs and laptops. The “up to $1000 PCs” you see in the Best Buys and Circuit City still have a lot of the low end Realteks in them that support up to 100 Full duplex only. And yes there is the OS throughput too like you mentioned. A HP laptop I bought last year at about $900 only has that capability. So with the OS throughput and interface factors in place, in the end it’s actually pointless even to have a cable with that speed.
The point I was trying to make is that it takes much more than just the cable. There are a lot that have the perception that you buy into a cable that can perform to that spec, you are going to get that speeds.
Yes, agreed… the speeds always start from the backbone and main core as it’s the one that govern the network.
I think there is a lot of life left in copper yet, but the real consideration is whether any of these new technologies require a replacement of the copper that is already there.
If they do (and I don’t know about cat-7), you might as well replace it with fibre. The Dr John invention from Melbourne though seems different as it promises to make what is already there much more able to reliably deliver the speeds we want in the future.
I wrote about this issue a few weeks ago, just after the news broke from Melbourne. Sorry for polluting your site with external links, but this is relevant!
Cheers
Jeremy
Wires are old school. Support Wimax.
Jeremy…ACK! Replacing in-building station copper for fiber is a huge mistake. POE? I’ve been in environments where the customer ran fiber to the desk and it’s nothing but trouble. No POE, the NIC’s are always expensive and the fiber is more delicate in general. What do you do about laptops?
Really bad idea.
Gerald, while Wimax might be cool if it ever gets deployed I don’t see enterprises moving to that for internal communications for quite some time. It’s been an uphill battle just getting them to accept WiFi and it’s never a replacement, rather an augmentation.
Yes, there may be life in copper BUT that Cat7 cable sure does sound expensive & bulky.
Fiber optic cable links are a reality and can be ruggedized much like copper cable. Eventually cost will dominate, and there will be cross-over to active optical cables (when is the question…?)
There is a LOT of pull for higher density cabling in the data center, and if you’ve seen the pictures of copper cabling
Component manufacturers are already developing 100 GigE offload NIC’s so when the wire or optical network connection is standardized, there should be a very fast adoption of 100 GigE by the HPC community.
Mike – we already have this product: http://www.grid-x.com