Dynamic Provisioning: the New Mantra

Dynamic Provisioning: the New Mantra – Technology Information

Andrew Feldman

In a commodity market, it’s all about value-added services.

Bandwidth is becoming a simple commodity and the downward price spiral is putting more and more pressure on service provider revenues. Providers must find new sources of strategic advantage and competitive differentiation to increase profitability and reduce customer churn.

New services, such as dynamic provisioning, QoS, classes of service, VPNs and managed firewalls, help service providers differentiate their offerings from the competition. Just as voice carriers found out years ago that new services like call waiting and voice mail helped them increase revenue per customer, higher-margin network services increase the average revenue per subscriber for today’s provider. Best of all, the cost to provide value-added services is nominal, but results in higher margins–just what service providers need.

In addition to offering value-added services, providers need the ability to provision those services dynamically. By combining Gigabit Ethernet’s massive bandwidth with the pinpoint control of dynamic, real-time provisioning, providers can offer customers the bandwidth and services they need, when they need them. Bandwidth on demand also allows providers to eliminate the over-provisioning common with static bandwidth allocation, thereby making the most of their networks.

Dynamic provisioning is the ability of service providers to adjust features and services to meet a customer’s rapidly changing needs. Traditionally when a customer wanted a new service or more bandwidth, the provider sent a technician to the customer’s site or to a remote location to reconfigure equipment manually. Truck rolls were costly and time-consuming. Dynamic provisioning is a far more cost-effective approach that allows the provider to turn up services or reallocate bandwidth remotely under software control.

By eliminating truck rolls, both the service provider and the customer win. Customers can increase or decrease bandwidth more quickly and request additional bandwidth for specific situations such as teleconferences. Dynamic provisioning also allows customers to order and receive services faster.

For the provider, dynamic provisioning speeds time to market and time to revenue. Dynamic provisioning improves pricing by carving bandwidth into slices that more closely correspond to individual customer needs and makes it possible to profit from all the bandwidth delivered. For example, if a customer pays for a single T3/E3 line, but typically uses only half of the 45-megabit bandwidth, the provider can sell the customer half the T3/E3 capacity and sell the unused bandwidth to another customer with similar needs. Since both customers value each bit of bandwidth they receive, the provider’s pricing will not suffer from the value dilution it previously incurred by distributing surplus bandwidth–at little or no extra charge–to a customer that didn’t need or want it.

Dynamic provisioning also reduces operational costs. Network operations expenses represent a majority of a service provider’s cost structure–far more than capital expenditures. According to some estimates, OAM&P (operations, accounting, management, and provisioning) expenses represent over 70 percent of total service provider costs. High OAM&P expenses further exacerbate problems faced by bandwidth-oriented providers because their expenses soar as their revenue base grows. Service-oriented providers, on the other hand, can deploy a variety of automatic provisioning tools–or even customer self-provisioning tools–to avoid costly truck rolls and dramatically reduce network operations costs.

Dynamic Provisioning

Service providers can use open-source APIs (application programming interfaces) to create Web-based, real-time service provisioning applications for features such as bandwidth on demand (see Figure 1). For example, a company holding a one-hour Web-based videoconference might need extra bandwidth for the conference. To obtain the additional bandwidth, the customer’s network manager uses the service provider’s Web-based self-provisioning interface to enter a request; within seconds, the bandwidth can be provided. When the teleconference is over, the bandwidth automatically scales back to the customer’s normal allocation.

Alternatively, a company network administrator can use a self-provisioning interface to turn up varying levels of bandwidth for individual applications, offices, or VPN connections. The applications made possible by such automated, bit-level provisioning are limited only by the service provider’s imagination.

Dynamic provisioning requires service providers to work closely with their infrastructure providers to ensure that the network infrastructure has the necessary functionality. The network equipment must offer bandwidth control and rapid service provisioning through bit-level rate limiting.

Hardware Rate Limiting

Service providers can implement dynamic provisioning through HRL (hardware rate limiting), whether by port or aggregate flow. Hardware rate limiting enables providers to pinpoint bandwidth control and provision bandwidth dynamically. Although rate limiting can sometimes place a very heavy burden on a router’s central processor, thereby reducing system performance, some manufacturers offer rate-enabled line cards to ensure that the system’s performance is not degraded.

PRL (port rate limiting) controls all available bandwidth on a physical port, turning a Fast Ethernet or Gigabit Ethernet connection into a lower-speed connection, based on the customer’s bandwidth requirements. PRL provides bidirectional rate limiting for service providers that want to restrict bandwidth on an inbound or outbound port, regardless of what is transmitted on it. ARL (aggregate rate limiting) provides a method to control bandwidth consumption on a specific traffic or protocol pattern, based on traffic policy. Using the same interface, service providers can use multiple traffic policies to create different limits for different types of traffic, thereby enabling complete control of the amount of bandwidth allocated to particular applications.

Both PRL and ARL use a hardware credit bucket scheme to provide bandwidth control on hardware. As the name implies, these credit buckets contain information about the amount of bandwidth (or other value-added services) for which a customer has paid, and regulates the provision of services on that basis. Hardware credit buckets are built into all rate-enabled line cards to ensure wire-speed performance. For PRL, the hardware credit bucket scheme limits how much traffic is allowed to enter or leave a physical port. For ARL, the hardware credit bucket limits how much traffic is allowed to enter an interface. The service provider controls the amount of credit in the bucket, depending on the amount of bandwidth allocated to the customer (the more bandwidth, the more credit for that type of traffic).

Making It Pay

It’s not a business if you can’t get paid, and you can’t get paid for dynamically provisioned services unless you can capture billing information. Traditional telecom carriers learned the hard way that voice accounting and billing systems are mission-critical. Failure to collect reliable account information can result in huge financial losses. Some providers will go so far as to shut down services if they lose the ability to track usage.

Regardless of the service, dynamic provisioning tools must be able to account and bill for allocations reliably, intelligently and in real time. These tools must also dynamically monitor the bandwidth to ensure that customers use only services for which they have paid. On-demand intelligence gathering capability distributed throughout the network allows providers to manage resources in real time.

Generating intelligence data throughout the network at wire speed sounds great. Unfortunately, a billing or customer-care application can be easily overwhelmed if too much data is presented. To prevent third-party applications from being blasted with irrelevant data, the provisioning tool must allow providers to select and record only the most important information. For example, a provider may want to engage full-detail accounting for a new customer to help the customer identify traffic patterns and behaviors. Once the customer is comfortable with the data and has negotiated a high-value contract with the provider, the provider can relax the data collection until it’s needed again. Flexibility is the key.

The transition from selling commodity bandwidth to selling services requires service providers to build more responsive and more accountable networks. Providers must be able to offer bandwidth on demand to meet the sporadic, high-bandwidth needs of business customers. But if they offer differentiated services, providers must monitor bandwidth usage to ensure that customers get what they have paid for, and then reliably account and bill for these allocations in real time. Service providers also need the intelligence to identify customers that consistently bump up against their limits for there lies the opportunity to sell additional services.

Andrew Feldman is the vice president of marketing at Riverstone Networks. He was previously senior director for worldwide product marketing at Cabletron Systems. Feldman received his MBA from Stanford University.

COPYRIGHT 2001 Horizon House Publications, Inc.

COPYRIGHT 2001 Gale Group