Next big pipe not there yet
Verizon tests 100 gigabit network using existing infrastructure
- By William Jackson
- Jan 07, 2008
IN NOVEMBER, Verizon successfully completed a 100 gigabits/sec field test, demonstrating it is possible to transmit 100 gigabits/sec traffic via a 500- kilometer optical link already carrying 10 gigabits/sec traffic, using existing infrastructure and commercial equipment.
Company officials called the test, which delivered a live video feed from Tampa to Miami, a milestone toward boosting network capacity to meet expected growth.
'The move from 40 gigabits/sec to 100 gigabits/sec will be exponentially quicker than the move from 10 gigabits/sec to 40 gigabits/sec,' said Fred Briggs, Verizon Business executive vice president of network operations and technology.
But we still are a couple of years away from upgrading networks to true 100-gigabits/sec capacity on a single channel without bundling channels, said Glenn Wellbrock, Verizon's director of backbone network design.
'We're probably going to need another iteration of modulation advances to get us there,' Wellbrock said.
The goal is to be able to move 100 gigabits/sec of traffic via a single channel for more than 1,000 kilometers and do it cost effectively.
Although the current tests provided the bandwidth on a single channel using Alcatel-Lucent's 1625 LambdaXtreme Transport system, it was not plug-and-play.
'It wasn't a plug-in card that would slide into the system,' Wellbrock said. 'It was a half rack of equipment' that used manual, rather than automated, voltage controls. 'Where we need to get is a plug-in module.'
He estimated that cost-effective equipment with double the polarization of current equipment would begin appearing in late 2009 or early 2010. By enabling four bits to be sent in each optical symbol with this equipment, 100 gigabits/sec traffic could be sent in a 25 GHz channel, well within the capacity of current electronics.
Boosting bandwidth into the tens of gigabits has required a fundamental shift in the way the optical signal is modulated. As bandwidth went to 2.5 gigabits/sec and increased to 10 gigabits/sec, bits were encoded by essentially turning the light on and off and measuring the difference as a 1 or a 0. 'We're doing something smarter with the light than just turning it on or off,' Wellbrock said. This is the technique used in the 100 gigabits/sec test.
When the advance was made to 40 gigabits/sec networking, the new technology was not understood particularly well, he said.
Because of the experience networks now have the move to 100 gigabits/sec should come about more quickly once the equipment is available.
It is not just video that is driving demand for increased bandwidth, Briggs said.
'Applications based on online video are clearly drivers for higher bandwidth, but there are many others from a large business customer perspective,' he said. 'For example, database applications and file delivery, as well as disaster recovery and business continuity solutions are also driving our customers' ' need for capacity.
Wellbrock said there will not be a wholesale upgrade to 100 gigabits/ sec even when practical equipment becomes available.
'Ten gigabits/sec is still more common today than 40,' he said.
'Forty is just being introduced.
Forty will probably be the standard in the future, and we'll introduce 100 slowly.'
It is more cost-effective to upgrade network links as the capacity is needed. Most connections for business customers today are no faster than 2.5 gigabits/sec, and 10 gigabits/sec trunks are adequate to serve them. As 10 gigabits/sec connections become more common, more 40 gigabits/sec trunks will be needed. Demand for 100 gigabits/ sec trunks will follow more slowly.
William Jackson is freelance writer and the author of the CyberEye blog.