QoS the optical way

To meet the ever-increasing bandwidth demand in today’s network, and the urgency to be able to pump more data into the pipes at higher speed, revolutionary developments have been made at the optical core as well as the metro area and the network edge.  

Recent advancements in dense wavelength division multiplexing (DWDM) technology have multiplied the bandwidth capacity to staggering limits by enabling lambda transport and switching capabilities. The next generation metro optical equipment provides bandwidth management by offering total protocol freedom and flexibility, and the ability to add/drop, cross-connect, route, groom and aggregate any type of traffic in any granularity.  

By eliminating the physical barriers of bandwidth limitations, we are now faced with another problem: How do we use this unlimited bandwidth to continue offering more variety of services with the same commitment to quality of service (QoS)? 

More bandwidth does not necessarily mean better engineering and better resource use.  In fact, it could mean more chaos and cause more traffic congestion if it is not optimally engineered and used.  To enable more bandwidth-intensive and broadband networking applications for users, it is necessary to also build-in the required intelligence to maximize the performance of the optical network.

The flexible bandwidth management scheme will satisfy any transition rate, no matter what a customer needs.

The Metro Evolution

Traditional networks were designed and optimized for a voice traffic.  As data traffic has increased, more devices have been introduced into the network at each layer: Sonet TDM at layer one for transport; ATM switches at layer two for traffic engineering; and IP routers at layer three for intelligent routing.  Each layer requires its own bandwidth management, making it almost impossible to provision across these layers in a timely and efficient fashion. Now the metro optical equipment is evolving with the goal to combine multi-layer networking into one highly efficient, multi-platform bandwidth management element. 

The intent is to provide flexible bandwidth management with TDM, ATM or IP traffic types at any increment. It can scale from 100% TDM to 100% ATM to 100% IP, or any incremental fraction in between.  The bandwidth manager is designed with built-in QoS support and can support multiple protocols without requiring hardware forklift upgrades as customers' traffic patterns shift from traditional voice to converged voice and data.

Recent studies of the telecom market (see Figure 1) verify the strong presence of the metro optical market as convergence toward optical networking continues.  As the transition from the legacy circuit switched networks to the next generation, data-oriented optical networks continues, metro optical equipment will provide the necessary tools to make this transition as smooth as possible. 

The flexible bandwidth management scheme will satisfy any transition rate, no matter what a customer needs. No matter how long it would take a service provider to migrate from an all-TDM to an all-IP network, a bandwidth manager offers the necessary flexibility and freedom.

QoS in the Optical Network

Even though pure optical networks are emerging, there continues to be a demand for all service types with no dominant protocol in sight.  To support legacy networks and also smooth the transition path from a Sonet ring-based network to an optical mesh network, support of multiple protocols is absolute paramount.  

Here is how bandwidth managers are positioning themselves to deal with the market transition and provide quality services along the way.

TDM

The Sonet (TDM) functionality of the bandwidth managers supports the quality transport of layer-one traffic. The nature of a circuit-switched TDM network is to support dedicated connections and real-time data transfer--and that’s what Sonet networks offer: 100% availability and reliability. The circuit-switched lines are still the best and most reliable for carrying voice traffic, therefore delivering the highest level of QoS. A bandwidth manager with Sonet STS and VT multiplexing and grooming offers guaranteed bandwidth and high level of service in the network.

ATM

The ATM functionality of a bandwidth manager supports QoS at the ATM layer independently or simultaneously with other layers.

To maximize the performance of data networks, ATM QoS technology has evolved to support multiple classes of service. ATM QoS defines five classes of service:

• Constant bit rate (CBR)

• Variable bit rate real-time (VBR-rt)

• Variable bit rate non-real-time (VBR-nrt)

• Unspecified bit rate (UBR)

• Available bit rate (ABR). 

The service class defines parameters and acceptable cell loss and delay variation through the network.  As each virtual connection (VC) in an ATM network is created, a specific service class is assigned to it. Each one of these virtual channels acts as a dedicated circuit to meet the level of services indicated in the service level agreements (SLAs).

Service providers are able to offer QoS-differentiated services based on the different service classes and varying bandwidth guarantees in the traffic contract.  Service classes and traffic contracts are enforced using input "policers" that enable bandwidth managers to measure input traffic rates and identify the violating traffic.

IP

As service providers are preparing to meet the increased customer demand for bandwidth with typical applications such as virtual private networks, voice over IP (VoIP) and Web hosting, the need for a new set of value-added IP services is increasing. These new services demand SLAs for multiple QoS classes.  Bandwidth managers can offer QoS at the IP layer for all transmission methods whether transmitting over leased lines, frame relay, ATM or even packet-over-Sonet links. They offer guaranteed QoS levels by SLAs to enable service providers to gain a competitive edge in their service offering.

To overcome the Internet’s best-effort performance limitations and ensure that mission-critical applications receive the required bandwidth and priority, traffic engineering and QoS support is needed.  Using multiprotocol label switching (MPLS) traffic engineering and QoS mechanisms provides the appropriate level of control and service guarantees that is needed to support today’s enterprise Internet. 

Traffic engineering optimizes network resource usage to maximize operational efficiency. To avoid under-usage of network resources to provide QoS, MPLS delivers two technologies to be used in conjunction, MPLS TE (traffic engineering) for operational efficiency and MPLS DiffServ (Differentiated services) for QoS guarantees.

MPLS TE works by setting up optimal paths across a mesh network to load balance among network routers and also provide failure recovery. This technique is similar to ATM virtual channels, or frame relay data link connections. MPLS TE creates a path through virtual tunnels that is monitored for optimum performance, avoiding congested and over-used areas.  But since MPLS TE has no concept of QoS, it cannot enforce the QoS policies. This is where MPLS DiffServ comes into play.

MPLS DiffServ works by setting a value in the experimental field of the MPLS label to enforce appropriate actions by the routers along the path.  This field prompts the router to either drop the packet or queue the packet according to its priority value. Therefore, MPLS DiffServ controls individual QoS queues.

MPLS-enabled infrastructures are ideal solutions for monitoring the end-to-end QoS, and enable the correlation of performance measurements data against the corresponding SLAs.           

Intelligent Optical Network Management Systems

With the abundance of available bandwidth, the time has come to concentrate on network management, service provisioning, service assurance and network survivability. How would a management system offset the bandwidth availability with smart service provisioning mechanisms to make the desired increment of the bandwidth ready at the customer’s fingertip according to the pre-determined (SLAs)? 

With recent advances in IP-based protocols such as MPLS, IP traffic engineering can be performed at the packet level without the assistance of transport layers.

The key to make the available bandwidth useful for the network is managing this bandwidth to provision and tear down circuits dynamically in a fast and economical way. That is why the new emphasis is on intelligent resource management. The new, intelligent optical management systems evaluate and manage service performance from the customer’s perspective, and are vendor and protocol agnostic. 

The systems use universal and widely accepted standard interfaces to provision services on-demand in a fraction of time that used to take the legacy systems to provision an end-to-end service. They provide automated monitoring, tracking, fault isolation and fault recovery mechanism for faster, root-cause analysis. They locate the more congested traffic areas and automatically reroute traffic through alternate dynamically created routes in a robust, efficient manner.

Imagine the savings in time and effort of building an end-to-end circuit, spanning across the optical and electrical network of devices, by eliminating multiple layers. If building a circuit would only require building the data packet layer and the optical layer, we can achieve simplicity and optimization in circuit provisioning by reducing layer provisioning. The concept is not too far from reality. With recent advances in IP-based protocols such as MPLS, IP traffic engineering can be performed at the packet level without the assistance of transport layers.

The next generation optical network will feature a standard signaling interface between electrical routing and optical switching that integrates the bandwidth provisioning process.  That’s why most bandwidth managers are adopting standard signaling interfaces to expedite bandwidth provisioning across the multivendor optical network. 

One such effort is the Optical Internetworking Forum (OIF).  The OIF has defined the optical UNI signaling interface that requires control messaging or signaling to communicate between the IP router and the optical network.

As proof that the industry is making progress in this area, at Supercomm 2001 in Atlanta, more than two dozens data and optical networking equipment vendors participated in a UNI interoperability demonstration. These devices used in-band GMPLS/UNI signaling interface to communicate.    
Sara Sedighi is Product Manager, Marketing, for Metro-Optix. She can be reached at sara.sedighi@metro-optix.com.

Visit Metro-Optix online.