OMEN
07-01-2010, 10:37 PM
Souping up the Internet for heavy-weight apps could 'transform the industry'
Computerworld - MIT researchers have developed technology that they say not only will make the Internet 100 to 1,000 times faster, but also could make high-speed data access a lot cheaper.
The trick to such dramatic performance gains lies within routers that direct traffic on the Internet, according to Vincent Chan, an electrical engineering and computer science professor at MIT, who led the research team. Chan told Computerworld that replacing electrical signals inside the routers with faster optical signals would make the Internet 100, if not 1,000 times faster, while also reducing the amount of energy it consumes.
What would the Internet be like if it ran that much faster?
Today, a user who has a hard time downloading a 100MB file would be able to easily send a 10GB file, with the Internet running 100 times faster, according to Chan.
"We're looking to the future when computer processors are much more powerful and we have much bigger downloads and applications," Chan said. "When we get more powerful processors, people will be clamoring for more speed. The question is, can these new processors and their powerful applications be supported over the Internet? Everyone will be using more high-rate applications, like 3D, interactive games, high-speed financial trading."
And when that happens, Chan said users of those large applications will run into choke points on the Internet. And that could happen as soon as 16-core processors hit the market, if not sooner. "I think the Internet will not be fast enough within three to five years," he added.
The answer, he said, is optical fibers, which carry light pulses.
Optical fibers are used widely on the Internet, spanning great distances and even continents. While they transmit information more efficiently than electrical signals, optical signals are complicated to deal with. A router, for instance, has problems handling optical signals coming from different directions at the same time. To get around that problem, routers on the Internet generally take in optical signals and convert them to electrical signals so they can be stored in memory until they can be processed, said MIT's report. After that, the electrical signals are converted back to optical signals so they can be sent back out.
That process eats up chunks of time and energy. Chan and his team have developed technology that would eliminate the need for such conversions.
Chan's architecture, which is called "flow switching," establishes a dedicated path across the network between locations that exchange large volumes of data -- from Silicon Valley to Boston, for instance. MIT explained that routers along that path would only accept signals coming from one direction and send them off in only one direction. Since the optical signals aren't coming from different directions, there's no need to convert them to electrical signals for storage in memory.
"If this can truly jack up Internet data speeds by 100 times, that would have a huge impact on the usability of the Net," said Dan Olds, an analyst at Gabriel Consulting Group Inc. "We'd see the era of 3D computing and fully immersive Internet experiences come much sooner.... If this turns out to be practical, it could be a very big step forward."
Dealing with network bottlenecks would be a huge accomplishment, said Rob Enderle, principal analyst at Enderle Group."Right now, the network is the bottleneck for hosted computing. This change could transform the industry as we know it," said Enderle. "We are going to need a faster Internet. We need it now. Currently, we only have about 20% [of available bandwidth] in many places."
Olds noted that there's no way to get around the fact that a faster Internet infrastructure will be needed to continue to push forward with better devices and applications.
"The Internet is going to have to become faster," he said. "We still have millions and millions of people added every year. But, just as importantly, consider the millions and millions of devices that are now vying for Net access -- all of the smartphones, sensors and other devices that need connectivity. All of these new users are generating new Web traffic and content at a furious pace, and the Internet needs to get faster to keep up."
Chan was quick to note that switching to optical fibers won't just mean better performance. It also will mean cheaper high-speed access.
"With bigger applications and more bottlenecks, you could buy extra bandwidth if you pay through the nose, but that's not something every user could do," Chan said. "Sure, you can increase the data rate, but it's expensive. With this new architecture, we can speed up the Internet but make high-speed access cheaper."
The MIT research team is testing the transport part of its architecture at Bell Labs in New Jersey, Chan said, and they're making sure the switch to all optical fibers won't cause any long-term effects on the Internet.
Chan, who is planning to start a company that will deploy and sell the technology, added that the next step will be to piggyback the new system onto some traditional networks in the U.S. in a limited trial.
"I think we have enough tests to know that the transport is ready and the architecture would work," he added.
Chan said he's been in talks with router companies about the new architecture. The router companies, along with major Internet service providers, would have to buy into his plan.
"Assuming the technology works as advertised, the main drawback will be the cost of rolling out the new gear and optimizing the Net so that it takes best advantage of the new stuff," Olds said. "New cutting-edge technology isn't cheap, and while prices will drop over time, they will only fall if the volume for the equipment is there."
Olds also noted that it might take a little doing for major ISPs and user organizations to swallow that expense.
"There will be considerable expense in adopting this new technology, not just in the new hardware, but in testing and optimizing the Internet so it can take advantage of the higher speeds," he added. "Given that the new equipment will be more expensive, it isn't going to be easy to convince the providers to adopt it right away. Just because the speed of light is the fastest thing we know, it doesn't mean they'll pay big bucks harnessing it so I can get my e-mail faster. They're going to need to be convinced that the higher transmission speeds and lower power requirements will help their bottom line."
Computerworld - MIT researchers have developed technology that they say not only will make the Internet 100 to 1,000 times faster, but also could make high-speed data access a lot cheaper.
The trick to such dramatic performance gains lies within routers that direct traffic on the Internet, according to Vincent Chan, an electrical engineering and computer science professor at MIT, who led the research team. Chan told Computerworld that replacing electrical signals inside the routers with faster optical signals would make the Internet 100, if not 1,000 times faster, while also reducing the amount of energy it consumes.
What would the Internet be like if it ran that much faster?
Today, a user who has a hard time downloading a 100MB file would be able to easily send a 10GB file, with the Internet running 100 times faster, according to Chan.
"We're looking to the future when computer processors are much more powerful and we have much bigger downloads and applications," Chan said. "When we get more powerful processors, people will be clamoring for more speed. The question is, can these new processors and their powerful applications be supported over the Internet? Everyone will be using more high-rate applications, like 3D, interactive games, high-speed financial trading."
And when that happens, Chan said users of those large applications will run into choke points on the Internet. And that could happen as soon as 16-core processors hit the market, if not sooner. "I think the Internet will not be fast enough within three to five years," he added.
The answer, he said, is optical fibers, which carry light pulses.
Optical fibers are used widely on the Internet, spanning great distances and even continents. While they transmit information more efficiently than electrical signals, optical signals are complicated to deal with. A router, for instance, has problems handling optical signals coming from different directions at the same time. To get around that problem, routers on the Internet generally take in optical signals and convert them to electrical signals so they can be stored in memory until they can be processed, said MIT's report. After that, the electrical signals are converted back to optical signals so they can be sent back out.
That process eats up chunks of time and energy. Chan and his team have developed technology that would eliminate the need for such conversions.
Chan's architecture, which is called "flow switching," establishes a dedicated path across the network between locations that exchange large volumes of data -- from Silicon Valley to Boston, for instance. MIT explained that routers along that path would only accept signals coming from one direction and send them off in only one direction. Since the optical signals aren't coming from different directions, there's no need to convert them to electrical signals for storage in memory.
"If this can truly jack up Internet data speeds by 100 times, that would have a huge impact on the usability of the Net," said Dan Olds, an analyst at Gabriel Consulting Group Inc. "We'd see the era of 3D computing and fully immersive Internet experiences come much sooner.... If this turns out to be practical, it could be a very big step forward."
Dealing with network bottlenecks would be a huge accomplishment, said Rob Enderle, principal analyst at Enderle Group."Right now, the network is the bottleneck for hosted computing. This change could transform the industry as we know it," said Enderle. "We are going to need a faster Internet. We need it now. Currently, we only have about 20% [of available bandwidth] in many places."
Olds noted that there's no way to get around the fact that a faster Internet infrastructure will be needed to continue to push forward with better devices and applications.
"The Internet is going to have to become faster," he said. "We still have millions and millions of people added every year. But, just as importantly, consider the millions and millions of devices that are now vying for Net access -- all of the smartphones, sensors and other devices that need connectivity. All of these new users are generating new Web traffic and content at a furious pace, and the Internet needs to get faster to keep up."
Chan was quick to note that switching to optical fibers won't just mean better performance. It also will mean cheaper high-speed access.
"With bigger applications and more bottlenecks, you could buy extra bandwidth if you pay through the nose, but that's not something every user could do," Chan said. "Sure, you can increase the data rate, but it's expensive. With this new architecture, we can speed up the Internet but make high-speed access cheaper."
The MIT research team is testing the transport part of its architecture at Bell Labs in New Jersey, Chan said, and they're making sure the switch to all optical fibers won't cause any long-term effects on the Internet.
Chan, who is planning to start a company that will deploy and sell the technology, added that the next step will be to piggyback the new system onto some traditional networks in the U.S. in a limited trial.
"I think we have enough tests to know that the transport is ready and the architecture would work," he added.
Chan said he's been in talks with router companies about the new architecture. The router companies, along with major Internet service providers, would have to buy into his plan.
"Assuming the technology works as advertised, the main drawback will be the cost of rolling out the new gear and optimizing the Net so that it takes best advantage of the new stuff," Olds said. "New cutting-edge technology isn't cheap, and while prices will drop over time, they will only fall if the volume for the equipment is there."
Olds also noted that it might take a little doing for major ISPs and user organizations to swallow that expense.
"There will be considerable expense in adopting this new technology, not just in the new hardware, but in testing and optimizing the Internet so it can take advantage of the higher speeds," he added. "Given that the new equipment will be more expensive, it isn't going to be easy to convince the providers to adopt it right away. Just because the speed of light is the fastest thing we know, it doesn't mean they'll pay big bucks harnessing it so I can get my e-mail faster. They're going to need to be convinced that the higher transmission speeds and lower power requirements will help their bottom line."