Advanced services will require providers to provision bandwidth on demand and to offer IP-based voice, video, and data. But is the technology ready?
PER LEMBRE, Dynarc
The new generation of Internet service providers (ISPs) and application service providers (ASPs) must attract and retain customers with advanced services that will require careful planning and quick provisioning. These services will place high demands on networking gear, and many providers will need to rethink their metro/access network solutions.
As this new paradigm unfolds, which technology will sit between the Internet-protocol (IP) payload and the optical infrastructure? Dense wavelength-division multiplexing (DWDM) will be present to solve the bandwidth issue. Syn chronous Optical Network (SONET) continues to advance, but the bandwidth offered by SONET can't be fine-tuned and the technology lacks the dynamics needed in service-optimized metropolitan areas. Asynchronous Transfer Mode adds too much overhead, doesn't scale well, and is expensive to manage in this application.
While the old infrastructure is based on offline trunk-circuit configurations and real-time packet routing, the new model will utilize a dynamic, restorable configuration layer that automatically allocates bandwidth to virtual switches. Provisioning will decrease from a cycle of months to a few seconds.
Access networks are undergoing a dramatic change as fiber deployment rapidly extends to business complexes, industrial parks, and residential multitenant buildings. Who could predict 10 years ago that fiber would be widespread today? Increasingly, service providers preempt the outdated copper infrastructure and drive fiber to more businesses than was thought likely as recently as a year ago.
Dark-fiber providers are deploying fiber in fine-meshed infrastructures in metropolitan areas at a great pace. In Scandinavia, there are currently dozens of new service providers taking advantage of the dark-fiber networks and deploying optical broadband access networks to corporate and residential customers. These networks are evolving today and will eventually replace the local copper loop.
The fiber-to-the-building (FTTB) carriers, not troubled by legal issues such as getting access to the local copper loop from the inherent provider, the PTT, or the local Bell company, are even outpacing digital-subscriber-line deployment. These carriers understand that copper cannot sufficiently deliver advanced services due to its bandwidth constraints.
Often, 10/100Base-TX is used to connect customers to the optical network because of its attractive price/performance ratio. When 100-Mbit/sec interfaces are used to connect customers to the optical network, the need for intelligent traffic shaping becomes obvious. The ability to determine the amount of traffic entering the network is the basis for planning the network. The use of standard T1 (1.544-Mbit/sec) or DS-3 (44.736-Mbit/sec) interfaces strictly limits the customer's consumption of bandwidth. These interfaces make it hard to provision new advanced services.
As the bandwidth bottleneck in the last mile is released by fiber, the next challenge is to enable bandwidth distribution to customers and for services based on real-time requirements. There is no doubt that new services will be IP-based. Optical access networks, therefore, not only need to facilitate bandwidth distribution, but these networks must also provide IP unicast and multicast routing at the edge to enable effortless provisioning of future services (see Figure 1). This functionality will allow access carriers to provide IP-based voice, video, and data services.
A new architecture, channelized reserved services (CRS) leverages the optical layer to enable advanced IP services on demand. Based on a scalable, time- and space-division multiplexing scheme, the channel technology is a modern, IP-centric version of time-division multiplexing circuits that supports both unicast and multicast traffic. Channels can be provisioned to match the bandwidth needs-at any bit rate-of an individual customer. This technology combined with IP addressing and routing allows service providers to cut provisioning time from months to seconds (see Figure 2).
Set up on-the-fly across the network, the channel technology enables traffic shaping at high connect rates. The channels are established in linear steps of 512 kbits/sec to offer jitter-free connections, which allow for delivery of quality voice and high-quality video services. Since there are no packet buffers within the channel, the delay is kept low and constant. Thus, the dynamic optical network can provide the quality of service required for advanced services.
There is more to it than efficient bandwidth distribution. The dynamic optical-networking gear automatically determines the bandwidth requirements and provides plug-and-play operation, leaving all management to the IP layer. IP-based traffic engineering such as resource reservation protocol and DiffServ is used to provide the hard resource allocations in the optical network. Multiprotocol Label Switching tags classify traffic and are used to decide the characteristics of a dynamic channel.
An "auto-provisioning" router switch combines IP routing and bandwidth channelization functions. It uses information from Layers 2, 3, or 4 to classify the traffic and to allocate channel resources according to the traffic needs. Service providers are deploying these router switches in networks in both Scandinavia and the United States.
The channelized reserved- services architecture uses a distributed switching and routing scheme, which allows multiple nodes to connect to a shared medium such as a ring or bus. Due to efficient spatial reuse of bandwidth, the distributed channel technology doesn't suffer from the bandwidth constraints usually associated with a shared-medium technology.
The centralized approach used in deploying CRS rings offers many advantages:
- Bandwidth can be upgraded in steps, adding more packet-forwarding capabilities by increasing the number of router switches on the ring.
- High-performance and lower-end rout er switches can co-exist in the same network.
- The centralized router switch does not need to be upgraded when adding another node to the network, reducing the amount of hardware by a factor of two.
- Dual counter-rotating rings offer fast recovery, within less than 50 msec.
The CRS architecture offers ISPs and competitive local-exchange carriers (CLECs) a network strategy that combines automatic bandwidth channelization and IP routing capabilities at the network edge.
Convergence is a key word in every networking vendor's vocabulary. A number of technological challenges remain, and economics will determine the speed of network integration and convergence. The need to simplify the infrastructure is tremendous. Eventually it will happen and advances in optical IP networking will meet the basic network requirements.
The emerging CLECs and incumbent service providers demand access switching and routing equipment, residential set-top boxes, and high-performance multimedia servers. Today, no single vendor can ship-or at least not package and attractively price-this equipment, because of the economical constraints in the access network. The vision is clear; however, next-generation service providers would like to offer advanced services to customers using a converged IP-based optical network.
That requires an empowered IP infrastructure, enabled by dedicating bandwidth to the streaming applications and to the on-demand service offerings, downstream as well as upstream. MPEG-2 (Motion Picture Experts Group) and high-definition TV signals can be carried over the IP network on dedicated multicast channels using multicast routing to enable users to connect to TV sessions.
For business access requirements, the tunneling of T1 traffic for voice is an important feature enabled by dynamic optical networks and will remain so for years. T1 circuits are packaged in IP packets and shipped onto dedicated channels, guaranteeing delay-free transport to the public-switched telephone network. The channel technology will connect at any bit rate, allowing service providers to effortlessly provision future services such as high-performance virtual private networks and high-quality teleconferencing (see Figure 3).
FTTB, both for residential and corporate customers, can release the bandwidth bottleneck in the local loop. The next hurdle is to intelligently distribute bandwidth on demand, which will allow service providers to automatically provision advanced IP services.
The CRS architecture empowers ISPs and CLECs by addressing automatic bandwidth channelization and IP routing in an efficient manner. The timing is right. Today, emerging carriers and service providers are seeking these capabilities. In the future, the evolution of optical IP networks will enable a converged service offering.
Per Lembre is a product manager at Dynarc (Kista, Sweden). The company's Website is www.dynarc.com.