New IP-based optical architecture for MANs
SONET/SDH, ATM, IP
A channelized link layer between Internet Protocol and fiber provides quality-of-service mechanisms, bandwidth management, and redundancy.
LLOYD GREEN, Dynarc
The metropolitan-area networking space has become the panacea of today's networking industry, and that's where battles will be fought and lost. The metro space is viewed as the bandwidth bottleneck between high-capacity backbone networks and broadband access services, which are poised for rapid growth. Recent reports by telecommunications market researcher RHK Inc. (San Francisco) predict the North American metro SONET market will reach $7.6 billion by 2004, up from $5.5 billion in 2000. Despite numerous predictions citing the SONET's demise, SONET is the transport mechanism of choice, with its inherent resiliency and protection schemes.
SONET/SDH was introduced primarily as a ring-based, high-speed, low-delay and reliable transport medium to carry public-switched telephone network traffic. However, the exponential growth of the Internet, private IP networks, and business-to-business e-commerce has rapidly changed the networking landscape. Bandwidth-hungry applications such as streaming video, multimedia, and videoconferencing are driving the need for a reliable and cost-effective transport mechanism that can automatically adapt to networking changes instantaneously without the need for manual intervention.
As the migration to packetized traffic evolves, the traditional SONET environment is not effective at transporting bursty, unpredictable traffic. SONET is far too static. On the other hand, Gigabit Ethernet and 10-Gigabit Ethernet are simple and provide high bandwidth at low cost. Ethernet was originally developed for enterprise environments, however. It is a best-effort service not capable of dealing with today's complex latency issues and bandwidth guarantees. A number of network equipment vendors have introduced next-generation SONET solutions that leverage SONET's strengths and, at the same time, overcome its weaknesses.
Today's SONET-based metro rings are neither optimized nor scalable for the demands of future IP network services. There is a need for an IP-centric networking technology that releases the full potential of fiber rings. The technology should offer optical performance and Ethernet simplicity coupled with resiliency and protection schemes of SONET rings. Several equipment vendors have developed products to address this vital market segment.
A new group of auto-provisioning routers was designed for use in a proprietary-architecture known as channelized reserved services (CRS). CRS makes it possible for service providers to rapidly provision advanced IP-based services on demand up to OC-192 (10-Gbit/sec) speeds and still maintain the quality-of-service (QoS) characteristics and restoration capabilities exhibited from SONET systems. It is an IP-based optical architecture based on a channel abstraction, which implements a thin link layer between IP and fiber. The channelized link layer provides critical functions missing in IP: QoS mechanisms, bandwidth management, and redundancy.
The explosion of Internet traffic on networks today has caused service providers and vendors alike to rethink their overall network strategy. Recent studies from RHK indicate IP traffic is growing at 200% per year. Internet connectivity is moving from dial-up to always-on connectivity, forcing providers to address problems never encountered before. These problems need to be addressed without severely changing the cost-economics end users are willing to pay.
IP is a simple and scalable protocol that has evolved from a primarily data-centric protocol to a full-service protocol capable of carrying voice, video, and multimedia traffic. Service providers now have a protocol capable of being application-agnostic; the dilemma is how to provide QoS for all these multiple services in an all-IP environment and still maintain service guarantees. Several approaches were tried and tested. The most common method is to overprovision bandwidth, which requires no intelligent bandwidth management schemes and is relatively simple to implement. But for those providers where bandwidth is at a premium, that is a costly proposition, which in turn had to be passed on to the end user. A new methodology is needed that can provide the necessary QoS guarantees, while at the same time intelligently managing bandwidth in a seamless and cost-effective manner.
There are two major demands that push the industry to introduce strict QoS in IP networks:
- To accommodate real-time services, there is a need for a technology that can reserve resources between the sender and receiver. Real-time applications need a smooth stream of data end-to-end. Voice traffic traditionally uses circuits to ensure that. When moving to the packet-oriented data world, there is a need to guarantee the quality of the network supporting the flow of data.
- Carriers and service providers should be able to assign and comply to strict service-level agreements with customers, guaranteeing certain service levels independent of the traffic fluctuations on the public infrastructure. To guarantee a service level implies control of network resources and the ability to log usage data, link capacity, and overall availability. As mission-critical applications are moved from the LAN to the WAN environment, quality agreements become even more crucial. Several industry initiatives have been launched to address these concerns: differentiated services, type of service, and Multiprotocol Label Switching tagging and various forms of traffic shaping and policy functions.
CRS architecture uses a channel technology and services-aware routers to provide QoS to any IP service. This channel technology leverages the optical layer to provide advanced IP services on demand (see Figure 1). CRS enables IP networks to use this channel technology to connect the service provider site with the customer at any bit rate. The combination of IP addressing and routing with the channel technology enables service providers to cut provisioning time from months to seconds.
Channels constitute a portion of bandwidth from one network location to another or to many other locations-the channels support both unicast and multicast traffic. Depending on requirements, the channels shape the optical network and can match the bandwidth need of an individual customer or the QoS need for next-generation advanced services.
The channel setup time is very fast-less than a millisecond. Channels are set up across the network, enable traffic-shaping at high connect rates, and offer jitter-free connections, allowing for delivery of toll-quality voice and high-quality video services. Thus, the dynamic optical network inherently supports QoS. Dynamic bandwidth allocation is also inherent within the CRS architecture, which allows bandwidth to automatically adapt to service fluctuations in real-time. The channelized medium has four major features:
- Multirate. A channel is a dynamic resource managed in quantum steps and up to the full capacity of the link. The bandwidth allocated to the channel can be changed dynamically during its lifetime.
- Fast establishment. The CRS architecture is designed to create channels quickly, due to the distributed architecture where resources are locally available at all nodes.
- Multicast. A channel can have multiple receivers, which enables true multicast operations (as well as unicast and broadcast).
- Simplex. Channels are simplex, making it possible to guarantee resources both upstream and downstream and thus provide inherent support for asymmetric traffic patterns with high-bandwidth utilization.
Channels are created locally at the sending node by allocating bandwidth over the segments of the ring necessary to reach the receiver(s). Packets are sent over a channel by addressing them to the receiver(s) and sending them using the allocated bandwidth. The receiver finally removes the packets from the ring (see Figure 2).
Service-optimized optical IP rings simplify a service provider's infrastructure. The CRS architecture uses a distributed switching and routing scheme, enabling multiple nodes to connect to a shared medium such as a ring or bus. Due to efficient spatial reuse of bandwidth, potentially functional portions of bandwidth in one segment of the network can be utilized in another part of the network; the distributed channel technology doesn't suffer from the bandwidth constraints usually associated with a shared-medium technology.
CRS rings enjoy many advantages from using a conventional, centralized approach. The bandwidth can be incrementally upgraded, adding more packet-forwarding capabilities by increasing the number of routers on the ring; high-performance routers and lower-performance routers can coexist in the same network. A costly centralized router does not need to be upgraded when adding another node to the network, reducing the amount of hardware. CRS offers protection and restoration capabilities while ensuring maximum throughput within the network domain. That is achieved by implementing dual counter-rotating rings. In the case of a link or a node failure on one of the rings, maximum throughput will be maintained on the other ring (see Figure 3). When that happens, the broken ring will automatically reconfigure itself into a bus, ensuring all undamaged segments of the ring will still be used to carry traffic. In the case of a dual link failure, the rings will automatically reconfigure into a dual bus. The self-healing mechanisms in the CRS architecture offer ring protection of 50 msec.
To take full advantage of the CRS architecture, the optical transport architecture is seamlessly integrated with traditional IP routing capabilities. The result is an optical "auto-provisioning" router, which integrates the functionality of an add/drop multiplexer, a switch, and an IP router. That reduces the number of network devices and thereby limits overall investment costs. The all-IP approach also enables the service provider to simplify maintenance and support. Due to the decreased number of network layers, management takes place only at the IP layer. Standard techniques are used for distributing routing and network policy information. Therefore, extra global addressing and complex routing protocols running in parallel with the well-known and accepted routing protocols for IP traffic are needed.
A new initiative has been launched to address the need of the MAN's lack of support for bursty, packetized IP traffic. As a result, the IEEE 802.17 working group is undertaking the creation of a new media-access-control layer standard for resilient packet rings (RPR). The working group will define an RPR protocol for use in local-, metropolitan-, and wide-area networks to transfer data packets at rates scalable to many gigabits per second. The CRS architecture is expected to implement these pre-standard specifications.
The most compelling benefit service providers realize through the deployment of an optical-network infrastructure is its ability to support a myriad of new revenue-generating, high-speed services quickly and cost-effectively from the optical domain. Service providers can drastically simplify networking elements, offer automatic provisioning of services, and create new business opportunities. The CRS architecture empowers service providers with the ability to dynamically assign bandwidth on demand as user requirements dictate, while providing the foundation for new advanced services with optical-networking equipment.
Lloyd Green is senior director of marketing for Dynarc (Sunnyvale, CA). He can be reached via the company's Website, www.dynarc.com.