Since next-generation services will require carriers to manage end-to-end services, not only within their own networks but on those of third-party content providers and partners, they will have to be able to orchestrate both off-net and on-net facilities. If they underestimate the initial complexity of managing carrier-class IP networks, the transition to next-gen services could be quite arduous.
More Services, Fewer Systems
The name of the game has become multi-service offerings, whether wireless, wireline, cable, broadband or video. Carriers will have to manage huge numbers of end-to-end components for voice, video and data, as well as add-ons like voice mail, LANs, CPE, routers and IP contact centers. Each component will fall into different trading models, such as wholesale and retail, each with separate systems for inventories and billing systems.
The result could be tremendous fragmentation, unless operators manage to converge their POTS-based ATM and frame relay components so that they can efficiently migrate to IP. This change will be possible only by consolidating networks onto a common infrastructure.
To create a common infrastructure, inventory management will be the cornerstone of OSS in an IP world. To others, it is a component that at least should not be underestimated.
"We expect as much as 80 percent of the development spend to be focused on consolidating OSS, as we need a standard operating environment that has a consistent look and feel of product and OSS—across multiple geographies, services and technology domains," says Glenn Presland, IP telephony product manager for VoIP OSS implementation at British Telecom. BT's initiative to move to a totally IP core must tackle the fact that the systems touched by inventory number in the hundreds.
"The dream is to have hundreds of billing, OSS and inventory systems in different areas converge to one," Presland says. "Then we can overcome multiple license fees [and] support contracts, and facilitate efforts for coordinated development when software upgrades take place—not to mention cutting expensive business-to-business interfaces among stove-piped systems." The inventory system has to manage all details about customers, data, configuration, platform and end-users. BT, thus far, has chosen MetaSolv to provide elements of VoIP, and Cramer for infrastructure and domain management related to auto-activation of complex IP equipment.
Despite a difference in size, smaller startup Covad expects the same trends as it moves to IP services. "It's a matter of degree, but the problems are similar," says Paul Grantham, VP for software engineering at Covad. Because many of its OSSs are homegrown, inventory changes were not documented as the company grew, so even though it has far fewer systems to consolidate than a Tier 1, Grantham must grapple with his own set of problems. "We're growing at 18 to 20 percent per month, so you can't help but lie awake at night thinking about how you can add systems, CPUs and scale while minimizing the disruption to architecture," he says.
Covad is in "high-integration" mode, as it is migrating 67 main applications. "We knew we were doing a heart transplant, but we didn't know how many arteries and veins were involved," says Grantham. Because inventory is right in the middle of everything, he says, "we end up finding lots of interfaces and undiscovered ways data is used, even though network engineers tried to predict everything that data service fulfillment needed." Grantham has been surprised by the scope of the applications that rely on the same pieces of information. "There are times that repair and customer relationship teams rely on the same data for specialized troubleshooting functions, or customer support through automated loop tests, for example," he says.
Inventory, fulfillment and customer relationship management are the triumvirate on which Grantham focuses. He turned up a CRM package from Chordiant, an inventory system from NetCracker and a new order management platform from Siebel.
"We knew inventory would be a bear to implement, simply because it touches so many facets of our business," Grantham says, ticking off the many pieces to be entered into a new system: DSLAMs, circuits, dark fiber, backhaul equipment from third-party suppliers with high-speed networks, local access equipment from LECs. "When we order the local loop from a LEC," he says, "and a loop ID from ordering is then aggregated through DSLAMs to ATM switches onto fiber optic channels bought from third parties, we end up with a lot of piece parts."
To reduce the number of piece parts in inventory, large deployments of trunk information record keeping systems (TIRKS)—responsible for bringing data and processes together—will have to co-exist with intelligent softswitches that truly understand networks and the business applications riding those networks.
"It's the ideal time to swap out systems, as PBX-related contracts are at the end of their cycles, and Centrex services are being evaluated as a migration path to VoIP," says Kevin Farrell, VP of product management at TeleGea. The company focuses on providing a multi-service platform through a hosted VoIP model to offer features like self-signup and number inventory management. "Carriers want to be up and running within 30 to 90 days with class-five functions like find-me-follow-me, three-way calling, distinctive rings, and presence, so core technology choices around feature servers and application servers have to be made quickly and incorporated into inventory dynamically and in real time."
Beyond migrating to IP, a service provider has several reasons to upgrade its current inventory systems. In the recent Stratecast Partners report "Inventory Management, Sector Assessment," which discusses factors for prioritizing inventory management, service providers said that managing CAPEX and OPEX were in fact most important.
Dynamic/Active Inventory
Inventory will play a much more active role in managing processes and data throughout IP architectures. Rather than existing simply as a database of record, inventory will be the very thing that drives other functions—analogous to the way mediation started as a passage pipe from switches to billing to rules-based systems, but has evolved now to an active system that pulls data elements from OSS and network elements.
Cramer founder and CEO Jon Craton asserts that inventory must dynamically manage the lifecycle of the network. "CRM evolved from being a customer database to actually managing customer lifecycles and providing business applications to support that management. Just as customers have a lifecycle to manage, so, too, will networks; so inventory must now take on a scale analogous to that of CRM systems," he says. "In the same way you would not build a CRM system by re-purposing a traditional customer database, you have to have the ability to build tailored applications."
That dynamic capability is necessary for capacity planning, which changes in IP because traffic engineering is not coupled as tightly to inventory. Carriers must have a truly "live" view of resources and routing patterns so that they understand capacity and do not overprovision.
Network capacity has to grow with customer orders, and linkages between self-provisioning systems and capacity management tools have to be aligned to create an up-to-date view. BT's Presland says that "just-in-time" provisioning is necessary: "You don't want a customer order to fail because there's not enough equipment, but you can't afford to overprovision bandwidth and infrastructure waiting for customer demand. They have to come, and then we will build—not the other way around."
"It's the notion of a dynamic system that reacts real-time and collects data in real time, so trouble tickets can be quickly assessed and tied back to the network, and customer impact monitored and proactively addressed," says Julie Wingerter, VP of strategy for NetCracker. She contends that to react dynamically, carriers should be tying inventory to trouble ticketing systems and customer databases. "In the past, problems with the switches or routers were discovered with trouble tickets," she says, "but now carriers should be able to ‘see as they go' through analysis by the inventory system, which should link to customer databases so that customers impacted by outages can be contacted proactively."
Although devising end-to-end inventory with IP involves plenty of complexity, in theory certain aspects could be simpler. "The inventory function changes significantly with IP, as the network itself becomes much more dynamic in the way services are established," says CoManage CTO Andy Fraley, noting that the network is in charge of routing and setting up the underlying service infrastructure. "With IP, there is no need for a 100 percent end-to-end service design; the access points and service types are all that is necessary from an inventory perspective."
Because operators do not have to design every last cross-connect for the hop in services across the IP network, the dynamic routing of the network to set up underlying service infrastructure mitigates some of the headaches from circuit-switched networks. "Where back-office design work for racks and switches accounted for about 50 percent of traditional inventory management, 10 percent of the facilities design function is [inherently] present in IP," says Fraley.
Theoretically, with auto-routing in IP, the network lays out its own connectivity, which requires more tightly managed capacity planning—a function that used to be handled completely by inventory in legacy environments, such as TIRKS or a loop facilities assignment and control system (LFACS). With IP, packets are routed over core network transport links, and traffic engineering functions configure IP networks and decide the route and transport links.
"Now dynamic capacity planning is necessary to see if and where the network is getting saturated. That relies on having a live view of the network so carriers know what services for what subscribers are routed over what core links," says Presland. He warns that the move to IP will not be fully automatic, as those wishing to enable self-provisioning and self-management would hope. "The technologies are not yet here for delivering multiple services over a common infrastructure, so provisioning that infrastructure in an automatic fashion is not yet possible," he says. "Fully automated self-provisioning and self-care products is what all operators want, but OSS is expensive enough that it could make operators' products too expensive for the market."
Of course, there exists "more than one way of skinning a cat," Presland says: "Carriers can purchase or merge with partners that possess MPLS network coverage in areas where they are lacking, or they can use dual vendors to 'gateway' among networks."
"The core tenets of OSS with MPLS networks is that configuration, provisioning and capacity management must come together to understand capacity constraints," says Adan Pope, Cramer's director of research. There's "limited understanding of how to do that in IP, which is traffic-engineered for setting capacity constraints on links to deterministically route paths."
Although the equipment providers say MPLS-based networks are self-managed, "provisioning of that infrastructure is very capacity-constrained and anything but automatic," Pope says. Carriers have to be wary of ending up back where they started. "You don't want to have to add network technologies and systems to manage and provision ATM, frame, IP, and all the normal stack requirements for provisioning for workflow, trouble ticketing and fault management all over again," he says. "Be very wary of recreating the sins of the past, such as over provisioning with big pipes." Rather, he suggests, make sure flexibility is at the core of systems decisions. "You want to be able to learn about capacity as the business grows, and provision MPLS IP cores according to that change."
"A great deal of thought and technical understanding has to be put into inventory before clear specifications for fully automating product activation and ordering and billing," Presland says. "To have all the bells and whistles for offering low-end products on the Web, you have to fully understand the operational pinch points."
From Stovepipes to Enterprise-Class Applications
Those pinch points can be eased in part by moving to converged, enterprise-class systems. Much as manufacturing, ERP and CRM companies apply design principles, with process and data coming together, IP service providers will have to break the habit of separating data into discreet systems and instead build horizontal, cross-domain systems that are scalable, flexible and configurable.
Some movement in that direction is evident already. Cramer, for one, is working with partners like SAP, Accenture, IBM and HP in the hopes of getting inventory to scale up to enterprise proportions, and MetaSolv was approved as a Cisco partner, which may lead to development of pre-integrated solutions.
"We absolutely hear over and over that carriers want more involvement from their EAI vendors, so we are looking at convergence and how we can get more value-add services so that digital content, enterprise mobility and value-add services are easier to implement," says Yancy Oshita, senior director of the global communications industry practice at Oracle. The organization is very active with the TeleManagement Forum and its New Generation OSS (NGOSS) initiative, and the Parlay Group through its Oracle 10g application server. "We think EAIs have to have true single data models so there can be integration on service delivery platforms and gateways," says Oshita.
For that idea to become reality, true pre-integration would have to take place. That takes tremendous investment, which has traditionally made it unpopular in OSS. "It's time that OSS suppliers worked in tandem with platform vendors and enterprise companies to establish a common understanding of operators' development roadmaps," says Presland. "Then they could cut down exhaustive testing of the products by the operators themselves."
Covad worked with Oracle and BEA Systems to implement high-performance, scalable systems. "We wanted to cut down reliance on proprietary technologies," says Grantham, noting that the company is looking at service oriented architecture with lightweight integration points—similar to BT's strategy—to decouple systems while growing.
"Implementing interfaces is the biggest challenge, and really, we should be able to buy those interfaces," says Grantham, who has "stirred things up a bit" by offering Covad's implementations as open source. "We hope the inventory and OSS vendors will embed interfaces into their offerings so that they can lower the cost of integration." He believes that "open standards promote business rather than constrain it."
The need for pre-integration may even drive OSS vendor consolidation, where standalone products that offered one function will now offer numerous functions. While inventory systems will remain distinct systems, they will have to become increasingly more flexible and tightly integrated. "Because auto-discovery is only possible with a limited number of network elements, and automated inventory management—soup-to-nuts—is dependent on auto-discovery, inventory management would be significantly enhanced by hardware vendor support of auto-discovery in their networks," says Lorien Pratt, program manager with Stratecast's OSS Competitive Strategies division and author of the "Inventory Management, Sector Assessment" report.
Stratecast research indicates that inventory management will break out into two segments. One will focus on horizontal solutions, which are strategic rather than tactical plays. "They will possess open architectures designed for OSS consolidation by putting multiple functions on single, extensible platforms that radically reduce total cost of ownership over time," explains Pratt. "Horizontal players have vertical plays, also, to solve certain points of pain, which they do often with partners or lightweight solutions on either end." She notes that vendors like Cramer, MetaSolv, Telcordia/Granite and NetCracker fit in this category, often focusing on Tier 1s, which seek longer-term ROI and have deeper pockets.
The other segment will consist of more vertically integrated offerings intended to expedite time to market for Tier 2s and startups, or organizations within Tier 1s. "These are short-term, tactical solutions that align around specified deployments—whether DSL, VPNs or others—that can be upgraded through time," Pratt says. Companies such as Viziqor, Axiom and HP (with its Service Activator) are players in this space.
According to Stratecast data, Telcordia/Granite has the largest market share, as its TIRKS has been in the back office for decades (see Figure 2) "Despite recent turbulence," says Pratt, "Telcordia has a big foot in the door—not only because of their plans to offer a converged Granite/TIRKS system, but because they have at least double [the market share] of their nearest competitor—Cramer."
She believes vertical and horizontal solutions will converge, and systems will handle integration of lines of business, as well as multiple transport, multiple customer types and geography.
As BT rolls out intelligent softswitches to deliver multiple services and consolidate OSS silos, Presland sees convergence "happening," but slowly, "as operators are having to play catch-up on OSS vendor choice as they look to manage products in the new converged world of data products, as they deliver voice and value-add services." He suggests that existing OSS systems, in many cases, will have to be "asset bled" for as long as possible to recover existing capital investment, and "that means new products have to be launched to meet the marketing window whilst OSS plays catch up." In Presland's view, COTS products falter when trying to accommodate rapidly changing markets and large customers' rapid changes and developments. "I have products that have specified feature sets, so when a large customer bid is won, there needs to be facilities and features to rapidly develop in a roadmap plan," he says.
BT currently is "scoping" an SME converged VoIP and data product, based upon broadband delivery. Essential to the product will be QoS on broadband access to ensure voice quality. Presland predicts the product will be available this year or next, "depending upon demand and the development costs to launch."
That type of choice reflects another trend toward a "market inversion," according to Pratt. "Rather than focus on specific services, carriers will focus on specific customer segments, thus blurring the lines among wireline, cable, broadband or wireless providers, and increasing the focus on consumer, SME and/or multinational segments," she says.
The bottom line is that carriers will want OSS vendors to react immediately when they redesign product catalogs and bundles. Vendors will have to demonstrate working OSS, rather than PowerPoint slides, and will have to constantly release software upgrades to expedite the selling of products in a competitive marketplace.
