Enabling the optical service networks with Multiprotocol Label Switching

May 1, 2000

Multiprotocol Label Switching will play a critical role in shaping and routing optical services.

Sean Welch
Tenor Networks Inc.

For service providers in the public networking arena, optical networking is the most important and profound technology since the introduction of Internet protocol (IP). While the optical service network of the future is an exciting prospect, there are significant technical challenges. Until recently, optical networks were enslaved by the time-division multiplexing (TDM) concepts of voice transport. Moving forward, optical networks will need to adhere to a much more flexible traffic-multiplexing concept; the new architectures must address the services that the optical network is expected to deliver.

An important technology requirement for the optical service network is the ability to carry voice and data and to work with multiple protocols such as Synchronous Optical Network (SONET), frame relay, Asynchronous Transfer Mode (ATM), and IP. Such a network would be based on a virtual infrastructure-no matter what physical media or devices are used. Providers could then offer any service, and those services would be constant, despite changes in the underlying technology.

Impossible as this may seem-it is not. The foundation for this concept was formulated by the Internet Engineering Task Force (IETF) in 1996 and has matured to become what is referred to today as Multiprotocol Label Switching (MPLS). MPLS is already playing an important role in the deployment of optical networks and evolving into a universal optical protocol.

The architectures that network service providers will deploy to build optical service networks will certainly vary. Despite the variations, these architectures generally consist of two distinct layers: the optical transport layer and the optical service layer-working in unison for service delivery.

The optical transport layer comprises optical transport switches (OTXs), which create and switch wavelengths (or lambdas, the Greek symbol used for wavelength) to form point-to-point connections in the optical domain. The OTX performs the following functions:

  • Lambda creation. The ability to add capacity at the optical level in the form of a wavelength that is then used to carry information from one point to another.
  • Light-path management. The OTX creates and manages point-to-point connections through the concentration of light paths through of a series of lambda-aware switches that are organized in various topologies. It also monitors the optical performance and power of the light path on an end-to-end basis.
  • Protection switching. The ability to provide an alternative route in the case of a physical failure in the fiber path or the equipment.

OTXs are the bedrock of the optical service network and used to form a very stable and resilient Layer 1 network. One shortcoming of OTXs, however, is that these devices are unable to make switching decisions based on the payload, or the information inside of the lambda, because the technology has not progressed to that point. Developing this capability is critical to unleashing the service potential of the optical network.

Due to technology limitations, switching intelligence still resides in the optical service layer. Using optical service switches (OSXs), lambdas are converted into the electrical domain where the payload is examined on a packet-by-packet basis and at the speed of light. These devices apply processing and switching intelligence to form more comprehensive services. Based on the addressing and control information found in the service payload, OSXs play a key role in directing the flow of content from one optical channel to the next. These switches work in concert with OTX equipment to apply the requisite service intelligence required to build robust service offerings, including the ability to offer rapid creation and activation, tiered services, legacy-data transport, and sophisticated measurement and accounting techniques.

MPLS is an important protocol that will help shape future services in the optically enabled public network. It can be applied in networks in different ways-to facilitate new business opportunities and solve technical issues. Many MPLS applications are emerging for optical networks; some are the subject of industry efforts in standards bodies and others are the focus of industry coalitions concerned with advancing the state of the network and promoting interoperability and reference testing.

Figure 1. Key applications driving the adoption of Multiprotocol Label Switching.

Vendors and analysts argue about whether MPLS is a packet-based or cell-based technology, not only because it can carry both types of information, but also because it works over both packet and cell facilities. This debate addresses the flexibility of MPLS, rather than its limitations. MPLS can do what ATM, frame relay, and IP does. Best of all, it can do it on electrical networks, optical networks, or a mixture of both and at speeds from kbits/sec to Gbits/sec.

Today, three key applications are leading to the widespread adoption of MPLS:

  • With MPLS, service providers can optimize traffic flows, resulting in better service to customers. In addition, MPLS in conjunction with IP differentiated services (diff-serv) allows service providers to deliver guaranteed quality of service (QoS) for real-time IP applications on an end-to-end basis.
  • MPLS recently found a role in optical networks, both for routing and for binding the optical network to the data network.
  • It also allows the universal adaptation of IP, TDM, ATM, and frame relay over a single network infrastructure.

These three applications are illustrated in Figure 1.

MPLS was first deployed in Internet service-provider networks for its ability to optimize resources across the network, thereby minimizing congestion. It performs traffic optimization by specifying the bandwidth needed between two routers or switches (such as an OSX), then finding a path through the network that has the available bandwidth. (Devices such as switches or routers in an MPLS network are generically referred to as label switch routers or LSRs). Based on service-level constraints, once all of the bandwidth along a path of predefined constraints has been reserved, an alternate path with sufficient bandwidth is found. For service providers, the result is improved resource utilization, control, and better service for customers. MPLS by itself does not support multiple classes of service, such as the guaranteed QoS needed for applications like videoconferencing.

Today, however, LSRs are starting to support diff-serv, which allows the marking of IP packets with the desired class of service. These classes vary from best effort, which is equivalent to the forwarding behavior of the current Internet, to expedited forwarding, which provides a leased-line QoS level. When diff-serv is coupled with the ability of MPLS to reserve bandwidth across a network, the result is a guaranteed QoS matched to the needs of the application.

Figure 2. The structure of the optical service network.

MPLS has recently been adapted to route light paths across optical networks. This application of MPLS-sometimes called multiprotocol lambda switching-is used by optical crossconnects (OXCs) to find paths with available lambdas between two endpoints. This application allows rapid creation of optical bandwidth as well as the ability to reroute paths after fiber cuts. It also provides stable private lines for use by voice or data networks.

In January, the Optical Domain Service Interconnect (ODSI) Consortium was formed to begin work on a reference model that will allow LSRs to signal OXCs to create light paths between LSRs on demand. This specification will enable data networks to rapidly adjust to different traffic levels by supplying additional bandwidth where and when it is needed.

The newest application of MPLS is as an "adaptation" layer that can support legacy TDM, ATM, and frame relay transport services (see Figure 2). In this scenario, the routing of the MPLS paths is similar to that of ATM and frame relay virtual circuits. Today, MPLS used in conjunction with the diff-serv framework, can deliver ATM-like QoS as it "adapts" traffic from either frame relay or ATM endpoint. Standardization efforts for TDM private lines, whereby voice traffic is transported over an MPLS network using concepts similar to performing circuit emulation over ATM, are just beginning. While this application of MPLS is still proprietary to a vendor's platform offerings, it is clearly an area of interest to service providers seeking to build new capabilities into the network.

The realization of the optical service network is eminent. The opportunity to develop and implement optical services with flexible QoS and bandwidth controls will allow service providers to differentiate their offerings and use bandwidth efficiently in a crowded marketplace. For example, an application service provider specializing in the data backup and recovery business could offer high-capacity connectivity to customers as it is needed. In the past, flexible capacity based on actual business requirements presented a formidable challenge. The vast capabilities of MPLS with its bandwidth optimization techniques make it a natural framework for the optically enabled public networks of the not-too-distant future.

Sean Welch is vice president of marketing at Tenor Networks Inc. (Acton, MA).