It's Not a Service Until You Can Bill for It

Mark Veil

In today's connection-oriented environments, service providers commonly apply usage-based billing to voice services and flat-rate billing to data services. In a connectionless, IP-based, broadband world, the requirements are no different - carriers need to be able to charge premium rates for premium services. This requires new tools to effectively quantify and measure a service's value, ensure contracted service levels are met and accurately bill subscribers.

Traffic classification and the ability to specify, capture, measure and account for appropriate quality and grade of service (QoS/GoS) metrics are critical to carriers looking to charge premium rates for premium services. New technologies are emerging that allow carriers to perform usage-based accounting at the connection, service and application levels. We will examine how carriers can leverage multi-protocol label switching (MPLS) and traffic management to deliver service-specific accounting in a connectionless, broadband environment.

Tiered network access, data and application services are a critical source of differentiation for service providers. In fact, the provisioning of raw bandwidth alone is no longer sufficient for new service providers to remain in business. The ability to offer a wide range of service products enables the provider to meet the "mass customization" needs of highly specialized markets while concurrently addressing the needs of broad, general markets. Of equal significance - at least from the provider's perspective - is the ability to considerably enhance revenue opportunity through strategic product differentiation.

Flat-Rate vs. Usage-Based Accounting and Billing The simple reality is that a "service" is not a "service" until its use can be billed to the customer. Flat-rate billing is relatively straightforward - requiring service providers to understand service activation and deactivation times and offered service level agreements (SLAs). Given this limited set of requirements, flat-rate billing may be effectively accomplished without the need to interact with network elements or operations support systems (OSSs).

On the other hand, differentiated offerings, such as tiered data and application services, require more sophisticated metrics to measure and account for the varying levels of value being delivered by the service provider. From a revenue standpoint, providers would also prefer to charge for network and service usage, not simply flat-rate access. In sharp contrast to "one-size-fits-all," usage-based billing requires knowledge of network resource utilization attributable to active services.

Billing to the Level of Service Customers are buying services, not bytes or traffic rates. Since this concept has significant ramifications, we will briefly define what we mean by "service." For our purposes, a service is defined by the application it is delivering, as well as the specific QoS requirements of that application - the particulars of which are provided by its SLA and traffic contract parameters. Moreover, a service is characterized as an identifiable flow, or subset of customer traffic, which transits a specific domain or end-to-end network path.

In other words, to accurately account and bill for services delivered, providers must have the appropriate element and network-level software tools capable of efficiently and accurately collecting service-specific utilization metrics from instrumented network elements. These network elements must also be capable of producing meaningful usage data at the appropriate level of granularity. In addition, network operators must have the ability to abstract accounting metrics at the level of the service from end-to-end perspective.

Connection-oriented technologies - such as time-division multiplexing (TDM), asynchronous transfer mode (ATM) and frame relay - inherently support this concept of end-to-end service level abstraction. Since each service is either physically or logically separated from one another, and the path of transit throughout the network is known, the matter of aggregating and correlating accounting metrics to the level of the service is a relatively straightforward affair.

Overcoming the IP Billing Challenge For connectionless IP-based networks, the crux of the challenge is that you can't bill for something that you can't effectively measure, which has significant impact on overall service level conformance monitoring and management.

Traditional IP networks operate on a "best-effort" basis, with all packets being individually routed to their ultimate destination on a hop-by-hop basis. Moreover, packets from the same flow will likely traverse the network over many different paths. Under these conditions, it is difficult to achieve more than the most rudimentary of accounting capabilities, let alone QoS guarantees. Generally, the only available metrics will be in the form of basic network-level interface statistics.

Traffic management services, such as packet classifiers and conditioners, may be employed within the network in order to differentiate between classes of traffic and ensure that prioritized traffic receives preferential treatment. The benefit is that accounting is moved from a network to a service level. While this provides a valuable new dimension to the accounting equation, these metrics must still be collected and aggregated on a hop-by-hop basis across the entire network.

In either of the above situations, the correlation of accounting metrics to the level of the service is nearly impossible, and it is certainly impractical. In such cases, service levels can be inferred, but without an explicitly identified route, or "connection," through the network - one that can be strictly managed, monitored and controlled - it is very difficult to deliver QoS, let alone account and bill for it.

Fortunately, an emerging technology called MPLS promises to deliver this missing link to IP networks and in doing so, provide an SLA and accounting management environment that is as feature-rich as today's TDM and ATM networks.

MPLS Overview MPLS was originally intended as a means to enhance routing performance. However, continued improvements have shifted the application focus of MPLS to its inherent capabilities for delivering efficient and scalable traffic engineering and QoS in IP-based networks. MPLS operates at layer two-and-a-half (L2.5) and is protocol agnostic to the layers above and below it. Its architecture is based upon the multilayer switch concept, which cleanly separates the forwarding and control functions; the former is under the domain of MPLS.

The power of MPLS stems from its ability to associate and allocate any type of user traffic with a particular forwarding equivalency class (FEC) (See Figure 1). Each FEC represents an aggregation of traffic that will be treated in the same manner as it traverses the network. Typically, these traffic classifications are based upon the contents of the packet header, such as the L2 and L3 source/destination address. In practice, however, classifications may be derived from (and applied to) a virtually unlimited range, combination and granularity of packet attributes - including physical ingress port/interface, application protocol type or Ipv4 type of service (ToS) and Ipv6 class of service (CoS) markings.

These FECs are then mapped to label switched paths (LSPs) that have been engineered to support specific traffic QoS requirements - guaranteed bandwidth or low latency for instance (See Figure 1). LSPs behave in a similar fashion to the more familiar ATM virtual circuit and frame-relay data link connection identifier (DLCI), but do so with much greater efficaciousness.

Upon ingress to a particular MPLS domain, all packets are assigned a label that serves to represent its FEC/LSP binding as well as shorthand reference to the contents of the IP packet header. In sharp contrast to today's "longest-match" routing paradigm, MPLS-equipped routers are able to perform ultrafast forwarding of IP packets via "exact match" label swapping. Moreover, MPLS overcomes the inherent limitations of traditional destination-based routing by supporting both explicit and constraint-based approaches for establishing LSPs. This capability allows network administrators to bypass potential points of congestion and direct traffic away from the default path selected by today's interior gateway protocol (IGP)-based networks and deliver precise control of network traffic and behavior.

Connection/Service-Level Accounting with MPLS The simplified topology drawing in (See Figure 2) illustrates the power of MPLS and LSPs for billing and accounting, and places the traffic classification illustration in context of a practical application.

In this example, an integrated communications provider (ICP) is delivering three basic services - each of which represents a specific FEC to the network - to a customer's corporate headquarters and a remote office:

• Voice service
• Premium data service with strict QoS requirements
• Basic data, best-effort service

The customer has seven basic traffic classifications - four data and three voice:

• D1: Data service that supports the corporate Intranet and priority LAN interconnect
• D2: Data service that supports a supply chain application delivered by the application service provider (ASP)
• D3/D4: Data services that support best-effort traffic such as TFTP and HTTP
• V5: Voice service that supports plain old telephone services (POTS)
• V6: Voice service that supports voice over IP (VoIP) services
• V7: Voice service that supports fax services

Upon ingress to the network, the traffic classifier will place the individual traffic flows into the outbound queue of the FEC/LSP pairing that matches its QoS requirements. Although there are only three levels of service being delivered to the customer by the ICP (again as represented by the three FECs, voice, premium data and basic data), it is understood that that each FEC will likely, at any given time, map to multiple LSPs depending upon the destination of the traffic. For instance, corporate Intranet/LAN interconnect and supply chain application traffic have been given the same QoS requirements and therefore are mapped by the classifier to the same FEC. However, since the destinations for the two traffic flows differ, each will be assigned a separate LSP.

It is important to note that LSPs support flow aggregation on a destination basis, irrespective of the flow source; if multiple network clients are accessing the ASP Supply Chain services, they will share a common LSP.

Conclusion Driven by the need to support QoS and service differentiation, the telecom industry is embracing MPLS as a means of delivering connection-oriented, deterministic behavior within today's connectionless, IP-based broadband networks. The primary benefit that MPLS brings to the accounting process for IP networks is the ability to elevate the quality and scope of usage data collection to levels typically associated with connection-oriented networks. Since all packets forwarded over a given connection travel the same path through the network, metrics can be collected at the level of the service, rather than just the network level. Correlating these connection-level metrics provides an extremely accurate picture of a service's end-to-end properties, ultimately enabling carrier-grade accounting, SLA conformance monitoring and service-specific, usage-based billing.

Top of Page