Network Convergence
As packet switching technologies rapidly overtake circuit switching and networks hybridize, the new rule is that whatever’s traversing the network is traffic, or content, without a real distinction between voice and data. Frame relay WANs are exploding, core switching networks are being built on ATM, and at layer 1 most traffic, packetized or not, is being transported over some kind of SONET or SDH infrastructure using wavelength division multiplexing (WDM). IP, PCM voice, circuit switches, SS7, ATM, frame relay, SONET/SDH, and WDM are all being combined to create the hybrid, global network. Voice may be carried on what are generally seen as data networks, but it’s really just another type of content transported over a hybrid infrastructure.
Convergent Access
The mix of network technologies must ultimately remain transparent to the end user. The key to this transparency is the access pipe. This is where the major players have focused their convergent service efforts to date. ATM is playing an increasingly important role in the access convergence market, as giants Sprint and AT&T have each unveiled aggressive, branded ATM access strategies-Integrated On-Demand Network (ION) and Integrated Network Connection (INC) respectively [DESIGN: PLEASE INSERT PAGE #(see The ATM Access Strategy, pg.XX)]. MCIWorldCom, with its On-Net brand, and other smaller, though significant, players, such as E.Spire with PLATINUM, are pitching channelized T1 offerings.
Technical details aside, these convergent offerings are largely marketing and pricing schemes designed to lock customers into long-term, exclusive relationships. One can buy T1 access from MCI WorldCom, Internet service from E.Spire, frame relay from AT&T, and long distance from Sprint all separately, but this is what these providers are trying to combat. They aren’t fighting it with fancy billing systems or complex rate plans, but with contracts that say, “The more you buy and the longer you commit, the less we’ll charge you for each service we already know you need.”
Pricing Schemes
These offerings diverge from traditional voice services pricing. Backbone providers are trying to break the inter-LATA/intra-LATA model of the PSTN and particularly the LEC networks. According to an MCI WorldCom spokesman, the company has created its own larger local calling areas and charges a flat rate for all local calls. The company also says that it uses point-to-point, flat, per-minute pricing for calls made across its network, and that the difference between calling from Dallas to Fort Worth and to London is in the range of a cent per minute. By keeping traffic on its own backbone globally, a service provider can avoid handoffs-which, according to the marketing pitches, means getting through to Copenhagen instead of the shawrma stand. In reality, however, a service provider that keeps traffic on its own network can avoid access charges along the way, and can thus drop its prices while maintaining a healthy margin.
As for connectivity and switching services such as frame relay and ATM, the pricing doesn’t really change. Prices are still based on the number of permanent virtual circuits (PVCs) and the quality of service (QoS) or committed information rate (CIR) associated with them. Internet access also remains at flat-rate pricing. Sprint was the only provider to offer a purely usage-based pricing strategy (see “Sprint’s New ION Service-A Major Test of Usage Based Pricing,” Billing World, February 1999). It has since backed off after finding that customers simply weren’t ready for billing by the bit. They wanted something more concrete on which to base their telecom budgets (see sidebar “Pricing for the Enterprise,” pg. XX[<
Billing Integration… or Not
Behind the scenes, at least for the major providers, the billing systems-the same ones they’ve been using all along-aren’t integrated yet. Frame relay billing systems bill for frame relay, ATM for ATM, long distance for long distance, access for access, and so on. The large service providers are using electronic stapling to send the end user a single bill and are not yet taking advantage of recent convergent billing systems in a broad, strategic sense. More often than not they apply such systems to specific products or applications, with the idea that if the system proves itself it could be a platform for a future convergent billing strategy. E.Spire announced in late 1998 that it built its back office on Kenan Systems’ Arbor/BP convergent billing and customer care platform. The single system supports all of PLATINUM as well E.Spire’s convergent offerings for small businesses. Of course, E.Spire was built from the ground up with convergent offerings in mind and wasn’t burdened with a large legacy environment. It does not, however, have as broad a range of wide-area connectivity services (such as frame relay and ATM) as the larger carriers, nor their global presence. PLATINUM is basically a channelized T1 access service combining local and long-distance voice and Internet access over the same pipe.
Integrated Customer Care?
MCI WorldCom, Sprint and AT&T are hammering away at not only the single bill, but also the single point of contact and single call resolution. In MCI WorldCom’s case, much as it runs multiple billing systems to support its products, it runs multiple support centers as well. If you call with a problem or an order, you’ll be thrown to an “intelligent” call router and sent to the person most qualified to handle your inquiry. If it isn’t the right person, you and information about your problem will be passed on to another center. The idea is to keep you from having to hang up and redial, by transferring you around and forwarding your information with what amounts to a dressed up, private e-mail system that links call center desktops. In AT&T’s case, there are centers the customer can call to talk to one agent with access to all of his information. This information is presented, however, in multiple systems-the “rolling chair syndrome.” Sprint’s situation is similar.
These companies do, however, have massive integration initiatives and hope to enable convergent customer views within a year or so. The next phase is to begin integrating the ordering and provisioning processes and offer end users self-provisioning. Wrapped in all of this is a focus on moving billing and reporting functionality to the Web, probably the most sophisticated value added for the customer today. The major players are beginning to use Web technologies to present billing data and allow analysts to “slice and dice” it in any number of ways. Call details can be broken up by region, service, department, time parameters, etc., to produce complex reports or downloaded for analysis with another application. Today the data is delivered according to standard monthly billing cycles. Real-time delivery appears to be a long way off and highly dependent on the aforementioned initiatives for back-office integration.
Convergence Drives Segmentation
Contrasting dramatically against the large carriers’ efforts to provide all-encompassing packages is some broad segmentation in how services are delivered. “A driver for success in this new market is going to be the ability to operate at different levels in the value chain-the chain being retail, wholesale, unbundled network, the ability to resell data, to provide it on your own network, or … take separate pieces of the network and resell those,” says Chris Bell, vice president for business development with Kanbay Inc.’s telecommunications industry group. The boundaries between backbone and access providers appear to be growing, and content provisioning may develop into yet another segment. This segmentation is likely to increase the need for both wholesale and retail billing systems.
Level 3 Communications, with its major fiber backbone across the United States, provides a clear example. The company is a carrier’s carrier, essentially, selling rack space in its points of presence (POPs) to access and content providers. Those partners serve end users and utilize Level 3’s backbone for transport. As they grow they can move into other Level 3 POPs, which are designed to be identical for a plug-and-play environment. The partners bill end users and pay Level 3 for transport.
Level 3 has also made deals with digital subscriber line (DSL) providers like Northpoint to allow its partners to resell access services. Conceptually, the company manning the rack in Level 3’s POP can just provide some value-added services over Northpoint’s DSL and Level 3’s backbone. In this example, the access pipe, transport and services may be branded together, but they are clearly separate entities with combinations of wholesale and retail billing relationships among them.
This segmentation is becoming increasingly apparent and logical. It’s apparent in the types of specialized carriers emerging, and it’s logical in the economies of scale created by one provider focusing on specific costs for bandwidth or equipment. The large carriers recognize this trend and its logic. MCI WorldCom, for example, has already invested $30 million in DSL provider Rhythms and will partner with others to gain high-bandwidth access where it lacks direct fiber links to customers. The goal, ultimately, is to avoid dependence on ILEC services.
Cost of Ownership In the convergence environment, the cost of billing system ownership becomes a major concern. As major carriers tend to use myriad billing systems, a large percentage of their total cost per subscriber-sometimes suggested to be as high as 70 percent-lies in the cost to bill. One of the selling points for a convergent billing system is the economy it theoretically creates by broadening the uses for a single system. If a carrier decides to move to a single system for such economies, cost of ownership considerations often turn into operating system decisions.
Companies using Microsoft’s Windows NT operating system often sell on the fact that their systems are cheaper to maintain, particularly compared with Unix-based systems. Thus far it’s been a price-performance tradeoff, with NT winning on price and Unix winning on performance. Wintel is reportedly due to release a new 64-bit chip, the IA-64 microprocessor, code-named “Merced,” which is supposed to add reliability and performance to NT. On the opposite side, Sun Microsystems’ telco OEM products group manager Jeff Veis said that Sun is aware of the demand for less expensive platforms and is working to reduce the cost of Unix ownership. At press time, Sun was close to announcing its new numbers.
Where Are Pricing and Billing Headed?
If it’s all just traffic on a common infrastructure, then how will pricing and revenue generation be defined? Many agree that data-style billing models are likely to become most common. “Billing will be more data-like than voice-like,” says Richard Klapman, INC product manager with AT&T. “We’re going to have common bills for all services and a common structure. If voice is transported as data, then cents per minute goes away over time. It’s not going to happen this year or next year, but directionally, cents per minute is not the way that data people think.”
This is not to say, however, that all-you-can-eat flat rates are the way of the future. Cash ultimately rests in what the end user values, and that’s most likely to be access, content, and service. One near certainty is that no one will provide an access pipe and transport for a flat rate without deriving revenue from what travels across them. The economics of flat-rate billing just don’t make sense. “Usage is a way to grow a revenue base. In a flat-rate scenario, once you get a customer it’s over. There’s no secondary or tertiary earnings stream there. From the carrier side, they realize that a monthly charge is much less lucrative than billing for usage,” says Kanbay’s Bell. For the customer, some flat rate models don’t make sense either. “I’m not sure a customer will be willing to pay, on a converged network carrying voice and data, the same amount for that voice call that they would for a video call,” says John Konczal, vice president for product management with Intertech Management Group. “And there’s no way the carrier will allow that video, which is more intensive on the network, to be priced like voice. The up-front costs to support that video are obviously far more expensive.”
Today, a compromise is emerging: tiered billing. VPN services, for example, are often priced by tier. Hours of usage are measured, and the total number at the end of the month determines which tier the user falls into. The more hours, the more one pays, but the less per hour. This relatively predictable model is important to enterprise telecom managers trying to work a budget. Bell says this approach is also a customer relationship management tool. Going with the old 80/20 rule, tiers create rewards for high-revenue customers. “You want to retain your best clients,” he says, “so you want to find ways to segment them and let them know in their bills that they are part of an elite group. It works as a reward mechanism and as a way for them to monitor usage. If they’re going to get certain discounts and certain tiers, it provides a clear picture of how [the reward] works.”
Discerning Traffic and Billing for Value
Today’s networks aren’t intelligent enough to distinguish types of traffic. A packet is a packet, a cell is a cell. Breaking through that wrapper to determine the content is an important next step. “I think more service providers would like to move to a value-based billing model. It makes more sense,” says John Yin, vice president for technology and product management with Daleen Technologies. “It’s the value that’s important. For certain applications, non-usage has better value, where for others usage or per transaction gives better value. That’s what’s driving this billing side. The service providers are trying to be competitive by trying to find that value.”
Value-based billing heralds a greater need for flexible billing systems that can alter rate plans and pricing schemes with relative ease. Much of the early modeling will be experimental, as service providers figure out just what users are willing to pay and test the limits of the demand curve. As Bell points out, “The overall question is the elasticity of the demand curve. For every drop in price, how sensitive is demand to that? Do you make it back in new volume?” The reverse is true as well: How high can a price go without limiting growth? This all remains to be seen, but it will drive the art of creative billing strategies and ignite furious price competition. Most of these battles will be fought on the enterprise battlefield, where the majority of today’s convergent service offerings are targeted.
Pricing for the Enterprise
Enterprise billing needs are creating some of the greatest challenges for service providers and billing vendors alike. Enterprise accounting practices are generally rigid and mostly standardized-a quarter is a quarter and a fiscal year is a fiscal year. Enterprises are budget-conscious, and their telecom managers seek predictability to keep the accountants happy. Very few telecom managers are interested in usage-based billing, says Richard Klapman, INC product manager with AT&T. “They want to have a fixed bill and a fixed price on a monthly basis because their budgets are fixed. But they also want usage statistics … to make sure the carrier is billing them correctly and end users aren’t abusing the usage. When you’re growing at, for example, 30 or 40 percent per year for frame relay and IP services, that’s something you really have to watch.”
Providing data that enterprises can use to forecast budgets is daunting. “This is a big problem because it can be a huge amount of traffic that needs to be tracked,” notes Konczal. He says he’s seen requirements where an enterprise wants individual records to be rated in order to determine a base price, and though the billing might be tiered or based on a sample method, the customer wants to know where the usage peaks fall. “That’s a lot of traffic to process,” he says. “This is very data- and storage-intensive. How do you architect the billing system? If you’re doing threshold or sample-based rating, when do you do that? Do you wait for a month-end billing close to determine the customer’s actual billed amount? Or do they want that determined throughout the month? If they want threshold billing, then the first X number of packets goes for one price, the next bunch for another. Do you want that in real time so that you have a true picture of what your usage is throughout the month?”
This issue becomes even more urgent as companies start to push for quarterly billing, Konczal says, and they want to project what they’re spending throughout the quarter. Standard methods for usage collection on packet networks are still being worked out, but it’s extremely complicated and processing-intensive to store all of the information, sample it, and determine rates. “It’s totally different from normal usage rating where I’ve got a call record, I know what it is, I rate it based on my current rate plan, apply it to the account, and at the end of the month give it a volume discount or free calls,” explains Konczal.
The ATM Access Strategy
It’s arguable that ATM access strategies are essentially an attempt to succeed where ISDN failed miserably. The concept behind them is what AT&T’s Richard Klapman calls “access economics.” AT&T’s INC, at least in phase one, will target the 700,000-plus branch offices in the United States that utilize separate T1 links for voice and data. The idea is to combine all traffic onto a single T1 with an ATM statistical multiplexer (MUX) on the customer premises.
After a site survey, AT&T turns up a parallel T1 to the customer’s existing services. The customer’s PBX is then dual-homed or rerouted into an ATM MUX along with the customer’s router traffic. The PBX interfaces to the MUX using a T1 or PRI serial line and PRI or channel-associated signaling (CAS). Once the link is tested and reliable, the other T1s are taken down.
Klapman says he took a lot of heat around AT&T when he proposed the use of ATM. “People said, “This is stupid, because IP is going to take over the world.” What’s happened since is that voice over IP is important from the desktop and on the Internet, but on physical access you want to go voice over ATM, and even the IP zealots are saying voice over ATM makes sense on access. We have a synergistic strategy where ATM is the transport core over SONET and DWDM, and IP is critically important to meet the customer’s application model.”
