How much TV is enough? Nowadays, it seems like we spend most of our time watching it—not sleeping or working or playing. Of course, that varies widely. But we’re all TV junkies. And while we have loads of TV now, we still want more. The affordability and availability of HDTV has only increased that desire.
We get our TV mainly by cable, but many still watch it over the air directly from local stations and increasingly by satellite. Now, Internet Protocol television (IPTV) is being delivered over high-speed Internet connections. While not widespread yet, IPTV is becoming available in more places.
Let’s not overlook YouTube downloads and uploads, which are clogging Internet service providers in many areas. They’re stealing bandwidth from uers doing more mundane things like e-mail and searches. All of this activity adds up to a massive overload of all our wired communications systems, from the Internet core to high-speed Internet connections and even to the cabling in our homes.
Though generally invisible to the public, the demand for ever higher data speeds is impacting the enterprise, especially in data storage. Many government and industry regulations mandate how data should be stored and handled, such as Sarbanes-Oxley and the Health Insurance Portability and Accountability Act.
Storage and archiving policies are forcing us to use greater amounts of storage, but that requires faster access and transmission times. There’s an overwhelming need for faster serial data networks, which are in the works.
Ethernet is the most widely used networking technology on the planet. Over the years, it has followed a regular upgrade path from 10 Mbits/s to 100 Mbits/s to 1 Gbit/s to today’s 10 Gbits/s. While 10-Gbit Ethernet (10GE) has been around for a while, it is just now being gradually adopted. Prices have dropped, 10GE switches are available, and there are more options than ever.
Specially enhanced versions of Ethernet like iWARP and other IETF-approved (Internet Engineering Task Force) extensions that use the remote direct memory access (RDMA) protocol mitigate the CPU usage and memory bottlenecks usually associated with TCP transfers. New iWARP network interface cards like those from NetEffect provide affordable 10GE performance in servers (see the figure).
As 10GE adoption continues to broaden, work is under way within the IEEE to extend Ethernet’s speed to the next order of magnitude. The IEEE’s 802.3 Higher Speed Study Group is working on a new standard called 802.3ba for 100 Gbits/s, or 100GE. It also will provide for an intermediate 40-Gbit/s version.
Designed as an interim solution, the 40-Gbit/s specification will take advantage of the technology developed for the Sonet/SDH OC-768 standard, which runs at 39.812 Gbits/s. The IEEE is considering a range of options, including 10 parallel 10GE fibers, 10 wavelengths (λ) of 10GE on a dense wavelength-division multiplexing (DWDM) single fiber, and four 25-Gbit/s wavelengths on a single fiber. It will be at least 2009 before such a standard or the hardware is ready.
The main problem with achieving 100 Gbits/s is the chromatic dispersion of the fiber, which lengthens the pulses over distance, thereby limiting speed. Considerable work has already been done in dispersion compensation, but those efforts must be extended to reach the 100-Gbit/s goal.
Electronic dispersion compensation (EDC) methods show promise, but higher-level modulation schemes like differential quadrature phase-shift keying (DQPSK) also are being considered. Few problems are expected at the usual Ethernet distance of 100 m, but plans call for 10- and 40-km versions that will require special compensation measures.
Already, 10GE is affecting other networking technologies. For example, Fibre Channel (FC), the most widely used storage-area networking (SAN) technology, appears to be a fading standard. This fiber-based network has kept pace with storage transmission speeds and is currently available in 1-, 2-, and 4-Gbit/s formats. Versions at 8 and 10 Gbits/s are in the works, but they may never be widely adopted because of their high cost.
FC is still the dominant SAN methodology, but its popularity for new installations is giving way to the Internet Small Computer Systems Interface or iSCSI (eye scuzzy) standard, which uses off-the-shelf and lower-cost Ethernet. FC is expected to decline further with the widespread adoption of 10GE and its application to iSCSI.
Fiber is the only viable high-speed medium for a future that demands very high speeds. In fact, it’s inevitable. With the plain-old telephone service and its aging twisted-pair topology in serious decline, major telecom carriers have no other choice.
AT&T and Verizon have recognized their gradually declining base of wired phone customers, as more customers adopt Voice over IP (VoIP) over cable or wireless as permanent alternatives. Given this trend, AT&T and Verizon have been gradually installing fiber to the home (FTTH).
FTTH is not new, just expensive. Fiber has always cost more than copper. Lately, though, it’s become more economical feasible thanks to passive optical network (PON) technology, which requires no active transceivers, repeaters, or other powered electronics along the path. With the speed demands of video and Internet connectivity in general, FTTH is seemingly developing into the only sane choice.
Verizon has been installing its FiOS PON system with its TV, VoIP, and Internet service in selected areas around the country—especially in areas with new construction, making installation faster and cheaper. Verizon expects to continue that investment into the future as funding is available.
AT&T is doing the same, starting with fiber to the neighborhood or curb (FTTC) and then using its twisted-pair phone lines in the last mile to the home with VDSL2 for high-speed connectivity. The primary justification for this network is AT&T’s U-verse IPTV service, which is designed to compete with cable TV. But along with that comes VoIP and really high-speed Internet connectivity. This rollout is small now but expected to grow over the years.
The Home-Networking Quandary
With more viewers and gadgets around the home, there is a real need to transmit video from the input cable or a PON to multiple rooms. Short connections still use cable, but how do you transmit digital HD video over hundreds of feet?
Home-networking systems have been developed to solve that problem and are a rapidly growing part of the infrastructure. While Wi-Fi wireless dominates home networking for PCs and laptops, there is some concern about its ability to transmit video reliably under the varied and severe signal propagation paths in homes.
The latest 802.11n draft 2.0 standard provides a base speed up to 300 Mbits/s, which is more than enough. But with severe indoor conditions, it may not be able to reach 100 Mbits/s or more, which is needed for multiple video streams over hundreds of feet. That’s why IPTV and satellite TV providers are looking to wired methods for distribution.
The Multimedia over Coax Alliance (MoCA) and the Home Phoneline Networking Alliance (HomePNA) are promoting the two systems with the greatest potential. MoCA uses the coax cable TV wiring that’s already in more than 80% of U.S. homes. Passive splitters divide the signal from the cable into streams for each room.
These splitters also can pass signals in reverse, making it possible for any node on the network to transmit in either direction. The system can deliver video data at up to 175 Mbits/s with quality of service, which is more than enough for current HDTV adoption levels. Verizon already has adopted the MoCA standard for its FiOS PON system. Comcast and Cox also plan to use MoCA in their new systems.
The HomePNA standard uses both the installed cable TV coax and the twisted-pair telephone wiring for its distribution system. It can achieve a rate up to 320 Mbits/s in its current form. AT&T has adopted it for its U-verse IPTV system. The MoCA and HomePNA systems seem to work well, and chip sets for both are available.
• Internet Protocol TV (IPTV) will continue to grow but will never reach the subscriber levels of cable TV. The related fiber will be appreciated for its super-high-speed Internet connection.
• The 100-Gbit/s Ethernet version will occur, but much later than expected due to technical difficulties.
• Fibre Channel will not die any time soon. It is too well entrenched for an early death. However, iSCSI will continue to take the bulk of new SAN networks.
• The power-line home-networking technology of Home-Plug and the Universal Power Alliance will miss the boat for IPTV home networking in the U.S., but will likely be a big hit in Europe and in some home monitoring and control applications.