Virtual circuits are poised to become more dynamic.
Unlike the "always on" permanent virtual circuits (PVCs) that have been prevalent in ATM networks to date, newer switched virtual circuits (SVCs) enable customers to set up connections in real time, thus paying for those connections only where and when they are needed. Although their adoption to date has been limited, SVCs have made significant strides and are a logical migration of PVCs.
PVCs define a path between network endpoints and are established by provisioning techniques performed within a service provider's back-office infrastructure. By definition, PVCs are "permanent" and offer continuously available paths across a network. Many users require guaranteed circuits 24 hours per day, and for these customers, PVCs are powerful solutions for their networking needs. However, PVCs might not be the best solutions for situations in which reserved bandwidth goes unused for significant periods of time.
SVCs, on the other hand, are ideal for service providers whose customers require connectivity only at certain times of the day.
Time Is Money
If Corporation A needs to transfer large amounts of data a few times each business day, it should not need to pay for a circuit around the clock. Using SVCs, the customer pays for the bandwidth established on a per call basis only during its peak business hours. The service provider can then utilize the bandwidth for other customers while Corporation A does not need the bandwidth. In such circumstances, service providers can accommodate more customers to maximize the infrastructure usage.
SVCs offer this type of service with connections established by signaling and user-defined endpoints. The circuit is created in real time when a user, switch or control signaling initiates the call request. A network with numerous PVC connections often requires meshed PVC networks and large amounts of unused bandwidth at any given moment; the same services provided via SVCs can be carried on a very simplified network on which customers share network resources, using them only when each needs the bandwidth. By taking advantage of statistical multiplexing, service providers can over-subscribe their circuits without compromising their customers' needs.
For the service provider, the equipment savings alone can be staggering. Meanwhile, the customer isn't forced to pay for unused bandwidth during idle hours. The customers can see their savings with detailed billing options, including usage-based billing, flat-rate billing and usage billing with built-in ceilings.
Beyond the financial savings, the potential for significant new revenue-generating services offered by SVC are significant.
New Services
Because of their flexibility, SVCs create the possibility for a variety of new applications.
One example is disaster recovery or backup services. In the event of a network disaster or lost connection, an SVC can be established to reroute traffic. For example, an SVC can support frame relay services with high availability requirements. If a frame relay connection should fail, the CPE could place an SVC call to a backup destination.
Network service delivery is another. With on-demand applications, SVCs provide flexible destination access. This allows cost-effective, temporary access to numerous destinations and simplifies network configuration.
A third is SLAs. With options such as usage-based billing, secure connectivity, and resiliency capabilities such as rerouting, service providers can offer their customers numerous SLA guarantees.
SVCs, of course, support any-to-any communications for video, voice and data. The flexibility with signaling destinations creates numerous options to meet customer needs. Voice, video and data calls can be made virtually anywhere, anytime, with high levels of QoS.
SVCs also provide support for services such as videoconferencing through dynamic bandwidth allocation. This creates efficient use of trunks and simplified administration, creating new revenue sources for service providers. Users also can save in areas such as travel budgets by utilizing teleconferencing capabilities.
With SVCs, users can take advantage of public network infrastructures for secure connections such as VPNs on an as-needed basis, often eliminating the need for private networks.
Quality and Security
SVCs now offer all of the QoS capabilities that private networks or PVCs provide. Advancements in automatic protection switching (APS), for example, can ensure that traffic that runs into potential failures along its route will be automatically switched to avoid interruptions or delays.
Because SVCs are dynamically established, it is essential that service providers be able to restrict SVCs to provide secure connections and guarantee paths for high priority traffic. They do this, in part, by configuring the "logical port" through which the signaling request for the SVC arrives. For example, service providers can restrict the bandwidth allocated to SVCs on a logical port. They also can limit the bandwidth available to each of the established QoS classes, and maximum values of allowed traffic parameters can be set for each service class. Service providers also can restrict the number of SVCs originating and terminating on a logical port.
Closed user groups provide networkwide security, allowing only those users that belong to the same "closed user group" to establish SVCs between themselves. A closed user group may be configured either to associate ingress and egress logical ports or to associate calling and called party addresses. They effectively implement VPNs in the ATM environment, thereby guaranteeing end-user security, even in SVC environments. Closed user groups operate in the core of the ATM network and extend to all multiservice applications from ATM to frame relay to IP.
In addition to the security features provided by closed user groups and the extensive logical port parameters for QoS guarantees, service providers can define security "screens" and calling party validation criteria and apply them to a single logical port. The security screens define address "block and accept" filters to be applied against the calling or called party in the ingress and/or egress directions. Calling party validation provides validation of the calling party signaled in the SVC request against the addresses that have been configured at the logical port of entry.
Further QoS guarantees are supported via priority routing. As with PVCs, a priority routing feature can ensure the highest priority circuits are established when there is contention for bandwidth. This can occur either when circuits are first established or when they are rerouted around failures. In either case, a switch can automatically route affected circuits to paths through the network that satisfy each circuit's QoS requirements.
Billing and Traffic Engineering
Of course, deploying SVCs is not without its challenges, but many obstacles are fading fast as development efforts continue to advance SVC networking capabilities.
For example, usage-based billing charges based on actual amounts of packets or frames sent requires switches to be tied in to back-office management systems for rating and invoicing based on call detail records. However, great strides have been made in network management solutions to tie these systems together, allowing service providers to take advantage of all that SVCs have to offer and map services to customer billing records and SLAs.
Similarly, traffic engineering concerns have been virtually eliminated. With ATM at the core and frame relay at the edge of so many networks today, service providers long have been interworking the two transports as outlined in Frame Relay Forum (FRF) standards. Original FRF standards were designed for PVCs, but new standards now include SVCs. Frame relay-to-ATM service interworking gives service providers the ability to extend SVC services in mixed environments while extending QoS across the two environments. In any network, the level of QoS is only as good as the weakest link in the chain. The extension of QoS to the frame relay switches at the edge of the network strengthens these links and provides the flexibility to go from the originating CPE to the destination CPE through the frame and ATM public network. This allows service providers to cater to very large installed bases of frame relay customers. Similar efforts are also enhancing traffic engineering capabilities in traditional routed IP networks as the multiprotocol label switching (MPLS) standard continues to develop within the Internet Engineering Task Force (www.ietf.org).
Clearly, SVCs have much to offer service providers and their customers. By enabling next-generation applications--such as SLAs and voice and video--and saving headaches and money, SVCs will continue to grow in popularity and be a force in switched networks of the future. The time has come for service providers to overlook some of the initial burdens of SVCs to take advantage of the competitive advantages they offer.
Paul Crann is director of product management for InterNetworking Systems at Lucent Technologies Inc. (www.lucent.com). He can be reached at (978) 952-7657 or pcrann@lucent.com.
