Google's recent announcement of a 1Gbps fiber-to-the-home testbed has communities across the US salivating—but imagine what the Internet might be like if that connection to your home were even faster. Say... 100Gbps. In less than 20 years, such speeds will be possible, but only for companies who installed the right sort of fiber architecture.
The UK telecoms regulator Ofcom commissioned a lengthy report on the future of fiber (PDF) (or "fibre," in this case) from the firm Analysys Mason. In it, the company sketched out the future of fiber capacity with a pair of handy charts. Both are clear: between 2025 and 2030, shared fiber tech will be able to offer 10Gpbs to each user; individual fiber can offer a full 100Gbps. Whether ISPs will support it or not is a separate question.
At the moment, fiber is divvied up into two main architectures. Passive optical networking (PON) aggregates data from a local neighborhood of users, multiplexing it into a single connection either by frequency division or wavelength division. Because they are shared, PON systems require less backhaul fiber and can be cheaper to install, but the total capacity of the fiber uplink is distributed between all the shared connections. (Note that this is the system used in the US by Verizon's FiOS, which currently uses a GPON architecture with a shared 2.5Gbps.)
The main alternative is point-to-point (PTP) Ethernet. Here, an individual connection runs from the home back to the local access point, providing the full capacity of the link to each user. This can cost more upfront due to the extra fiber and ducting needed (the costs to do trenching and home connections remain about the same), but it can also provide much faster speeds.
With that in mind, take a look at the charts below. The chart on top shows the expected deployment schedule for PON technologies, which break into time division (TDM) and wavelength division (WDM) just after 2010. The chart on the bottom shows the expected deployment speeds for PTP Ethernet.
In both cases, speeds are ramping up fast. PON technology is expected to offer 10Gbps downloads in the next year or two, though this will amount to a guaranteed per-user capacity of "only" 160-320Mbps. Verizon is already trialing the XG-PON technology in the field, announcing in December 2009 that it had done its first successful 10Gbps test outside the lab. Standards for XG-PON should be finalized this year, but it won't be in wide use for some time. (Verizon called it a "technology validation" rather than a "product trial.")
Somewhere around 2027, PON should offer 100Gbps of download capacity; split between end users, this should guarantee 10Gbps downstream connections.
PTP Ethernet will also offer 100Gbps downloads in the same timeframe, though each user should be able to see these speeds. While the infrastructure in the ground won't need constant updating, the equipment at the customer's home and the network aggregation points will. Moving to wavelength division systems means, for instance, that the optical gear in your home can transmit on different wavelengths of light and that the ISP's gear can demux all those wavelengths at the other end.
The report notes, "we do not believe that there will be any significant barriers in the evolution of FTTH systems capacity." Perhaps not from a technical side, but will US ISPs actually roll out such technology? And if they do, will we be able to afford it?
Then, of course, there's the use question: beyond streaming 1080p uncompressed films over the Internet, what would one do with all that bandwidth? If the Internet has shown us anything thus far, though, it's that people always find a use for more speed.
Source: ars technica
