There has been some recent talk of IP and Ancor. The June issue of Telecommunications gives a glimpse of IP's importance and future.
telecoms-mag.com
IP on Glass: Goodbye ATM & SONET
Public Networks
The all-IP network architecture allows IP bits to be placed directly on fiber for high-speed transport to their destinations, eliminating intermediate protocols, simplifying the architecture, lowering costs, increasing reliability and reducing errors.
Stan Hanks
The new public network is upon us with the advent of high-speed backbone networks that transmit nothing but IP bits. This new breed of network is revolutionary because it delivers the type of all-IP
architecture that has been described conceptually for several years, and now has the capacity to provide services well into the future. Here is a network capable of transmitting 1.44 trillion IP bits per second on each fiber route. Even with the massive bandwidth demand of today's IP networks, this capacity greatly overshoots present market needs.
Now, the ISPs, competitive and incumbent local exchange carriers (CLECs and ILECs) and others providing IP-based services will be able to secure the high-capacity, long-distance broadband capabilities
they need to meet the bandwidth requirements of their high-speed xDSL, cable modem and wireless subscribers. Because this new type of network is IP end-to-end, there is no need for special protocol
translations or gateway functions. The customer simply connects via high-speed IP access to the backbone and terminates in the same manner at the destination end.
The network architecture is so simple that its real value might not be fully appreciated without a comparison to existing network architectures. The all-IP smart architecture provides lower delays than
TDM networks. All the non-IP packet, frame and cell protocols between the IP layer and the physical circuit are eliminated. The "IP on glass" catch-phrase clearly describes this architecture, which
allows IP bits to be placed directly on fiber for high-speed transport to their destination. This eliminates intermediate protocols, simplifies the architecture, lowers costs, increases reliability and reduces errors.
Simple Architecture
IP bits entering the network have two options. The first is for IP bits to be sent across a circuit, which, due to the circuit's dedicated nature, provides an excellent quality of service (QoS) guarantee, but at a very high cost. The second choice is for the IP bits to be sent across the ATM switching fabric, which provides an excellent guarantee of service and the ability to support variable bit rate data. At a cost of 10.5 percent to more than 60 percent of overhead, plus ATM equipment and personnel, however, the true cost of ATM, even in a bandwidth-rich network, is often thought to be too high. By contrast, IP bits enter the IP network and are placed directly onto logical dense wavelength division multiplexing (DWDM) channels, each operating at a rate of 2.5 billion bits per second by high-speed backbone routers.
In these networks, high-speed backbone routers decide which path the IP bits will take across the backbone, and DWDM creates the high-speed optical circuits. DWDM logical fibers operate over a fiber
network. There is no need for ATM, frame relay or SONET.
The IP on glass architecture is a straightforward one: IP bits come in, they are transported across a high-speed circuit connection provided by DWDM, and then IP bits go out. The customer is responsible for
providing the actual interface to the end user and for doing whatever preparation work is needed, depending on the user's actual interface requirements. Voice over IP will have to be employed for telephony applications, video processing must already have been done for video over IP and conferencing applications, and data applications will have wrapped their information in IP protocol bits prior to the bits entering the network.
The router receives the IP bits from the customer IP connection, makes a fast (under 800 nanosecond) hardware-based forwarding decision, and rapidly sends IP bits to their destinations. Customer IP
connections can be set at 45-Mbps (DS-3), 155-Mbps (OC-3c) and 622-Mbps (OC-12c) levels. A customer-side router may be provided if the situation necessitates one.
After leaving the router, the IP bits move over 2.5-Gbps optical networking interfaces (ONIs), which are OC-48c-like interface connections between the router and the DWDM system. From the DWDM system, signals are optically multiplexed into one of dozens of available wavelengths of light for transmission to a terminating DWDM system. At the receiving end, functions are performed in reverse order so that the customer's IP bits may be sent to their destination.
All Layer 2 and Layer 1 overhead has been removed and bits are transferred directly to a timed optical circuit operating at Layer 0. Layer 0, the photonic layer, is another aspect of this network implementation that has long eluded network architects and is also a sign of future trends. Layer 0 is a layer below the traditional Layer 1, the physical layer of the OSI Model. Layer 0 provides a transport for traditional Layer 1 bits such as SONET, if needed, or otherwise provides transport for higher layer bits.
The all-IP backbone completes prior architectures by providing an IP-only core. Other layered, multitiered architectures of this type rely on ATM for a multipurpose, multimedia type of transport. The simple
IP-only approach allows customers the freedom to determine the architecture for distribution and access tier operation that best suits their needs. IP service-specific carriers can go IP all the way from the
customer desktop to the edge, through the core and back out the other side. Carrier customers might choose to use an ATM technology for a multiservice distribution tier, but strip off the ATM before providing pure IP to the transport layer. In either case, a variety of traditional or emerging access technologies may be employed--from ISDN and frame relay to ATM residential broadband access to cable modems and xDSL devices. This type of architecture provides customers with the ultimate in IP-only flexibility.
As the network architecture shifts away from traditional circuit-switched TDM telephony, an all-IP backbone can provide the flexibility that carrier customers require. The IP-only aspect of the network backbone provides an open architecture in which customers are free to implement the specific gateway types and manufacturers they require. There is no interface difficulty if the output of the gateway or translation device is IP, and customers can translate the number and range of traditional signaling protocol elements or special calling features they desire.
Carrier customers using an IP-only backbone may use any standard or proprietary scheme for voice encoding, conferencing, video encoding or encryption. This allows the backbone to supply raw bandwidth for transport only or to be a geographically diverse part of a carrier-customer virtual private network (VPN) solution, thereby providing the answer to bandwidth-starved carriers' main question:
"How do I get there from here?"
In so many ways, this emerging network design and implementation represents a glimpse of the future. From a technology standpoint this type of network offers a simple, yet powerful, high-speed,
high-capacity broadband network infrastructure that enables carrier customers to realize their dreams of ubiquitous, fast IP bits. The architecture, which top carriers had predicted is three years away, is
here today.
Stan Hanks is vice president of research and technology at Enron Communications. He was responsible for the invention of the GRE tunneling protocol and deployment of the first carrier IP
VPN in 1991, development of the metropolitan area Layer Two facility that grew into MAE East, and has been involved in many ISP start-up efforts. Hanks graduated from Rice University with degrees in computer science and electrical engineering and spent five years in Rice's Ph.D. program in computer science. He is currently active in the Optical Internetwork Forum, USENIX and CAIDA. Contact him at stan_hanks@enron.com. |
