4 Comments

Summary:

When it comes to capacity on the long haul cables under the sea or surrounding cities, there’s always a need for more bandwidth. Which is why Ciena’s announcement of a chip that can deliver 17.6 terabits per second over existing fiber cables is so amazing.

iStock_000009182804XSmall

You know what the Internet is going to need? More bandwidth. Because, really, when it comes to capacity on those long-haul cables under the sea or surrounding cities, there’s always a need for more bits per second. This is why Ciena’s announcement of a chip that can deliver 17.6 terabits per second over existing fiber cables is so amazing.

The company’s new chip that can provide speeds up to 400 Gbps on existing fiber and double the capacity of those cables from the current 8.8 Tbps to the 17.6 number. It says the same chip can provide 100 Gbps speeds over the 9,000-kilometer trans-Pacific cables connecting the West Coast and Japan. That’s a lot of YouTube videos.

But like some over-excited infomercial host, Ciena is promising even more. This chip, which should begin delivering traffic on networks before the end of this year, is also programmable via an API. So even if the speeds aren’t impressive (they are), the potential for programmable access networks is. Much like programmable networks inside the data center are a game-changer, being able to adjust capacity on a fiberoptic cable via software, as opposed to swapping out telecommunications gear, means operators have new options for their products and pricing.

Gather round, it’s time for some physics.

When it comes to fiber — or any communications technology, really — the challenge for engineers are the tradeoffs between capacity, latency and distance. Think of it as the more packets you shove into the pipe, the slower they will travel (capacity vs. latency). Additionally, the more packets crammed into the pipe or the faster you want them to get someplace, the shorter the distance they will be able to travel.

The big trend in fiber is to use electronics to come up with better modulation schemes that can continue pushing capacity, distance and latency so that even as engineers have to compromise, they are at least compromising from a better baseline. And now, by adding programmability, Ciena is letting operators compromise on the same piece of fiber using software. So if a client wants to allocate a chunk of the pipe for low-latency traffic that doesn’t transmit a lot of data it can, without having to change the hardware.

Ciena isn’t the only company pushing the limits with fiber chips. Alcatel-Lucent, Cisco, Infinera, Huawei and many others are hoping to profit from our bandwidth demands. Next week in LA there is an entire show dedicated to those feeding our fiber networks, so I expect more news and more insights on how fast we’re going to get to folks deploying 400 Gbps pipes. After all, it was only last year that Verizon deployed 100 Gbps pipes in the U.S. and in Europe. How soon will we need more?

You’re subscribed! If you like, you can update your settings

  1. Pretty sweet. Do you think this will have any practical effects on the throttling arguments currently being pushed around by ISPs?

    1. Not significantly. User demand for bandwidth is growing at 100% every two years (driven largely by Moore’s law), while spectral efficiency of optical links is growing at a fraction of that. Coherent systems, such as the Ciena transponder will provide a means of meeting a few more years of capacity growth but by the end of the decade a radical new approach to optical transmission is required to prevent a global bandwidth crunch – several groups around the world are working on this at the moment.

  2. Now if I could only get fiber to my house. Goodbye Comcast. :-)

  3. Richard Bennett Friday, March 2, 2012

    The average Internet connect speed has actually declined for the past two or three years as more people spend more of their Internet time using smartphones, so the “always needs more bandwidth” thing isn’t really true.

    These chips that speed up the undersea cables need to be installed in every undersea repeater, right, and not just at the landing spots? Gee, how much does that cost?

Comments have been disabled for this post