Robert,
Yes, I think so too. Cisco's announcement is significant, although it represents a natural outgrowth of the technology, IMO.
For more on this, see below. It is from this month's Lightwave magazine at:
pennwell.com
[I couldn't copy the specific url for some reason, so you can find the entire article posted below my blurb. Note: the hard copy of Lighwave containes some good illustrations that are not available at their site.. I guess they'll get their additional share of subscriptions this way... ]
The article implies throughout some very powerful realities to come, as far as the fate of some carriers are concerned, far beyond the face value of the printed words.
I don't agree fully with all of the assertions cited within the article that lead one to believe in the inevitability of IP in _all_ aspects of transport (although recent developments in router/routing architectures have caused me to re-evaluate some old notions and positions toward some otherwise time-sensitive (isochronous) applications), but in the main, Shariff makes a whole bunch of good points, probably in furtherance (but not yet deliverance) to my religious conversion.
Equally as important, and probably more so, there seems to be the critical mass of requisite public sentiment behind Cisco and its initiatives in this area, and sometimes, for better or for worse, that is all that matters.
Enjoy,
Frank Coluccio
----------------------
[APRIL '98 LIGHTWAVE MAGAZINE ARTICLE:]
Packet Over SONET Fuels New IP Transport Paradigm
by Michael Shariff, product manager for Packet over SONET and Gigabit Switch Routers at Cisco Systems Inc., San Jose, CA.
The Telecommunications Act of 1996 aims to eliminate the artificial distinction between inter-local access and transport area (LATA) and intra-lata business by opening the long-distance and local markets to competition. This deregulation of the industry creates regional competition. As a result, incumbent local exchange carriers (ILECs) will need to provide differentiated services to maintain and expand their existing customer base. ILECs will also need to penetrate the markets in which their customers conduct most of their business. In this newly formed competitive environment, the ability to increase revenue and earnings will determine who will survive. As a result, LECs will no longer be just local.
Yet with the local calling market estimated at $100 billion, stakes are high and competition is fierce in the ILEC's home market. According to the consulting firm Deloitte & Touche, 36% of business customers are ready to walk away from their existing ILEC to such new service providers as competitive local exchange carriers (CLECs), competitive access providers (CAPs), out-of-region LECs, and cable-TV carriers. The ILECs stand to lose as much as $10 billion in annual revenue from this customer defection. Meanwhile, CLECs are aggressively expanding to bid for this market. They now have more than one million lines in service and an annualized run rate of $1 billion in local revenues.
At the same time that the local and regional market becomes more crowded, business customers have begun to ask for differentiated treatment for their mission-critical services. Throwing bandwidth at this problem is no longer an economical or viable choice in today's competitive market. Thus, just as service providers are going through revolutionary changes, their networks and the nature of their services are going through an evolutionary phase as well. This creates an opportunity for those service providers who play an aggressive role in this evolution and who are early adapters of new technologies and services to survive this tide of change. An example of such an opportunity is the swing toward networks based on the Internet protocol (IP), and the emergence of a technology that makes the construction of such networks efficient and economical.
IP-centric networks
The demand for bandwidth is skyrocketing in data-service networks. With this demand comes the need for efficient bandwidth utilization, higher performance, and simplicity. As the rate of data provision begins to exceed that of the traditional voice services, the current time-division multiplexing (TDM) infrastructure becomes increasingly inefficient. The edge of the network belongs to Ethernet, requiring that networks of the future be designed for this protocol. Also in today's networks, transmission control protocol/World Wide Web traffic constitutes 75% of all backbone traffic.
According to The Yankee Group, Boston, the telecommunications market will grow from $195 billion in annual revenue to $260 billion by the year 2000, with data services accounting for the lion's share of this increase. Figure 1 shows the sharp rise in data services against the moderate increase of voice services in future networks. As the time line is expanded, the growth in data services by 2005 will be 23 times that of voice services. This increase has already been seen in the construction of data infrastructure, which is outpacing that of voice 3 to 1.
In this context, the explosion of Internet traffic has changed the network traffic landscape. Traffic types are changing from text and e-mail to images and video. These changes affect the characteristics of the packets traversing the Internet backbone. The average packet size has increased to 280 and 330 bytes this year alone. In addition, the average session duration is only 13 sec long. These factors point to exhaustive signaling connections that can overburden a connection-oriented network.
The traditional telephone company network structure is also being shaken by increasing interest in voice-over-IP (VOIP) technology. Business customers spend millions of dollars on international voice and fax traffic. The advent of VOIP means that these customers can spend a fraction of today's cost for voice and fax services. Internet telephony will compete effectively against the circuit-switched infrastructure, forcing the incumbent service providers to partner or build up their IP-based infrastructure before the new service providers carve away at the main source of their income. As VOIP technology evolves and the number of Web users grows to 160 million by the year 2000, more businesses will opt to put their long-distance transactions on IP-based networks rather than on existing circuit-based ones.
These trends have brought about the birth of a new paradigm: IP transport networks. The IP transport infrastructure is an IP-based, full-service network that takes advantage of IP Layer 3 quality of service and such inherent IP characteristics as scalability, simplicity, and multicasting. Today's fast-growing service providers are investing billions of dollars to build global long-distance and local infrastructure services via an IP-based fiber network.
Most of these networks are built from an IP transport vision with a focus on coexisting with the carrier's existing Synchronous Optical Network (SONET) and Synchronous Digital Hierarchy (SDH) infrastructures. sonet/sdh technology is the technology of choice worldwide for providing high-capacity pipes. SONET networks provide high resiliency and availability in the face of network failures and transport valued synchronization bits that can be easily tapped into for maintaining a synchronous data network. Today's data network can build on these features and take it a few steps further by adding better use of wide area network bandwidth, support for differentiated services through Layer 3 QoS, and high-performance multicasting features.
Packet over SONET fuels the IP transport infrastructure
Packet-over-SONET technology can relieve today's network from the stress of growth while paving the way for additional service features in the future. Packet over SONET/SDH is the serial transmission of data over SONET/SDH frames through the use of point-to-point protocol (PPP). This data mapping is done in accordance with RFC 1619, PPP over SONET/SDH, RFC 1662, PPP in high-level data-link control-like framing, and OC-3/STM-1 (155-Mbit/sec), oc-12/stm-4 (622-Mbit/sec), and OC-48/STM-16 (2.5-Gbit/sec) rates.
The format of IP packets in the SONET synchronous payload envelope is simple. One byte is dedicated to the packet flag at the beginning of the string and four bytes are dedicated to the PPP header (one address byte, one control byte and two for protocol bytes). The IP packets are serially placed in the SONET frame, and a cyclic redundancy check is added at the end.
Clearly, the routers that form an IP transport infrastructure that can coexist with the SONET infrastructure must access various overhead bytes in the section, line, and path SONET overhead. Again, most data platforms must be able to access these bytes in order to provide a cohesive network to the service providers.
Today's competitive market leaves no room for wasting precious resources. Deployment of packet-over-sonet technology will provide up to 98% efficient wide area network bandwidth utilization in the network.
Depending on the packet size, the ATM cell tax can create major deficiencies in the bandwidth usage in the network. The cell tax range could range from 14% to 50% of the payload. This becomes even more critical on long fiber runs, such as cross-country or transoceanic links.
At OC-3 rates, the effective data bandwidth for packet over sonet is 149.76 Mbits/sec versus that of ATM, which is 128.36 Mbits/sec. Small packets grind hard on an ATM network.
Summary
Networks must have longevity and scalability with an initial low cost of ownership. A successful network must also use bandwidth efficiently. To achieve these goals, we must design networks based on the type of traffic it carries and choose between a connection-oriented and a connectionless network early in the design process. Choices will depend on the applications running on the network. If an IP-based network is selected because the true nature of the infrastructure is data-centric, then ample buffering must be seeded in the network to deal with the round-trip delays and packet bursts.
Packet-over-SONET technology implemented with new gigabit switch routers can meet these requirements. Consideration of IP over SONET when a business model is IP-centric is essential. Packet over SONET leverages existing SONET technology and delivers simple and efficient bandwidth use, Layer 3 QoS, and scalability. It can synchronize off existing SONET rings and support the required bytes in the SONET overhead for reliable, cohesive integration.
Michael Shariff is product manager for Packet over SONET and Gigabit Switch Routers at Cisco Systems Inc., San Jose, CA. |
