Coming Down the OSS Turnpike
Cutting the Integration Tax

by John Williamson

Here's a question: What might the following activities have in common? Checking out at a Wal-Mart store, buying McDonald's take-out and ordering a broadband digital subscriber line (DSL) connection.

A possible answer: On average, these are all transactions consumers require to be completed in around 90 seconds, reveals research cited by Granite Systems, service resource management specialists based in Manchester, New Hampshire.

Current telecommunications industry wisdom has it that the reason why use of DSL hasn't as of yet become as commonplace as everyday shopping or fast-food consumption is in large part due to a failure of the technology's associated operations support systems (OSSs). In short, failure to deliver what is otherwise a proven technology.

DSL isn't alone in this. As networks continue their tortuous transition from highly deterministic time division multiplexing (TDM) circuit supremacy to highly indeterministic Internet protocol packet dominance, gaps in the repertoire of OSSs are now threatening the successful mass-market deployment of many new telecom and Internet services.

This OSS challenge -- some might say headache -- is very real to all pedigrees of player.

Established carriers will likely have big networks with multiple legacy systems, and inflexible proprietary OSS software, points out Trudy Nyden, vice president of marketing at outsourcing IP service provider Peak XV, San Ramon, California. "That's an OSS black hole that doesn't seem to have a bottom," she says.

Conversely, newcomer service providers might lack OSS expertise and, argues Pan-European optical network operator Interoute Telecommunications, they might not be being well serviced by the OSS supply industry.

"2001 is the year of OSS. New carriers are, however, being hindered by the lack of a full suite of functions and by the cost of OSS and business support system (BSS) components," maintains Ted Rook, the operator's vice president for network design. "There's an awful lot of hype about, mainly spread by people with a lack of knowledge or a vested interest." London-based Interoute says it is constructing what will be one of Europe's largest telecom infrastructures.

Then again, you find IP-oriented enterprises such as Qwest acquiring asynchronous transfer mode (ATM) oriented enterprises such as US West, notes Sayeed Azhar, Chelmsford, Massachusetts-based senior product manager for MPLS and QOS at Cisco Systems' IOS Technologies division. Presumably such hybrid service providers have complex hybrid OSS problems, too.

The ideal, according to service support software concern Trendium Inc., is to be able to dynamically manage at, and assure, the service level across different network layers and technologies. "Most management software today still manages networks and not services," says Claude Hary, Trendium's EMEA (Europe, Middle East and Africa) general manager and vice president, based in Sophia-Antipolis, France.

In this context, legacy networks generally get bad press. This might not be at all justified. "Legacy networks haven't disappeared -- they are here and even growing. Actually to call a network -- the PSTN, SS7 or SDH which brings the service provider 95 percent of revenues -- 'legacy' is almost insulting," says Avichai Levy, vice president of marketing at TTI Telecom International, a network management and OSS concern based in Petach Tikva, Israel.

Nevertheless, it's accepted that managing the transition from old to new is a real tough nut to crack.

"Are service providers going to migrate data? When are they going to scrap their old systems? Can they integrate systems to talk to each other?" questions Julie Wingerter, vice president of strategy at Netcracker Technology, a Web-based inventory management and provisioning concern based in Waltham, Massachusetts.

The migration task has sometimes been compared to changing the engines on an aircraft when it's still in flight, only this time there's the added vexation that the pilot doesn't quite know where the flight is supposed to touch down.

"This upgrade presents a number of challenges to providers as they now need to address, one, the migration of OSSs with sometimes millions of users riding on those systems/data and, two, how to introduce a significant amount of new capability into their systems to future-proof them for the next few years, even when they are not themselves sure where and what they'll need in services and capabilities," says Brenda Toonders, vice president of product marketing at Atreus Systems. Atreus is a broadband service creation platform concern in Santa Clara, California.

The upshot of all of the above, according to Ultan Kelly, the Dublin-based director of marketing for Marconi's network management and OSS business, is that "some operators actually now view their OSSs as being more important than the specific devices that are deployed in the network."

Worlds Apart

A major issue for the OSS community is that the IP world is technically very different from, and in many particulars much more complex than, the TDM world. This is apparent on several levels.

For starters, there's a basic difference of technical focus. "One of the main differences is that TDM networks are primarily focused on delivering Layer 1 services very quickly through DSxN or OCxN services. Being able to provide quick provisioning times, restoration and infinite bandwidth scale are critical," says Robert Travis, director of software services marketing at optical networking concern Sycamore Networks. "IP networks are focused on Layer 3-plus services -- quality of service (QOS) and class of service (COS)." Sycamore is headquartered in Chelmsford, Massachusetts.

Also, the comparable requirements in various parts of the OSS spectrum can be quite different.

Provisioning is one example. "IP systems are more complex to provision, particularly with respect to billing. IP needs the notion of a connection in order to be effectively managed," reasons Mike Linehan, director of product management and marketing at Integral Access, a multiservice IP access vendor, also located in Chelmsford, Massachusetts. Integral reckons to accomplish this through the use of multiprotocol label switching (MPLS) in its Purepacket operations management system.

IP billing is an item of common concern to fixed and mobile operators."There are no established standards for the records and tickets you can retrieve from the [mobile] IP network elements," says Dr. Jens Trotscher, vice president of product strategy at the Dreieich, Germany, office of IT and business services company Sema Telecoms. "There's a market for specialized companies that basically create, filter, aggregate and correlate these types of records."

Otherwise, though, there are some rudimentary differences in the OSS purview of fixed and mobile networks. "In fixed networks, the dominant engine for changes to configuration is customer orders. So we're serving the engineering and infrastructure and network operations user, and the customer-provisioning user," comments John Borden Jr., president and CEO of Granite Systems. "In mobile there is no change to the network configuration when a new subscriber is added or when a subscriber changes service parameters. The network is immutable in that sense."

Hyde and Seek

Meantime, in the manner of Dr. Jekyll and Mr. Hyde, service-level agreements (SLAs) and COS norms are acquiring different personalities in the move from circuit to packet networks.

Sure, an IP SLA can be set up to be oriented to issues similar to those expressed in frame relay SLAs — for example, bandwidth, burst capability and, to a certain extent, loss tolerance. "In all cases, these performance management attributes complement or masquerade for the goals of committed throughput and response time," points out Jean Hammond, founder and chief strategy officer of Quarry Technologies, an IP services edge switch concern based in Burlington, Massachusetts. "Enforcing predictable performance is now possible in IP networks."

IP SLAs also can express several different capabilities such as access control and variable billing rates, or other service attributes such as response time and reliability or restoration rate, Hammond adds. SLAs in IP-based systems also can be classified by technology, application or any number of other combinations, she says.

Interoute provides a case in point, using MPLS to prioritize traffic at the application level. "Because MPLS enables the identification and prioritization of different types of data, carriers using this technology have the potential to offer guarantees that are far superior to the simple 'uptime'guarantees common at the moment," says Matthew Finnie, Interoute's vice president of products and services. Interoute says it is now moving toward offering application-based SLAs, enabling it to prioritize SAP or Oracle traffic, as an example.

COS attributes also are different in an IP landscape. Integral Access notes that in the circuit-switched world, each class of service gets its own circuit -- its own connection or, in the best case, its own nailed-up time slot or slots.

By contrast, in the IP world, services share a common access link with multiple services running over the same link. In this case, bandwidth has to be dynamically allocated among active services, with the most time-sensitive and critical services getting priority.

"Consequently, in an IP world, differentiated services must be identified at the packet level since allocating a separate access link or nailing up DS-0xN is not possible," asserts Integral's Linehan. "To do this, real-time, mission-critical and best-effort traffic types must be identified -- by packet markings, ingress ports, IP addresses, UDP [user datagram protocol] address and so on — and forwarded on a path that can deliver that packet with its class-appropriate QOS." Linehan believes MPLS is the most efficient way to do this.

Technology aside, the telecommunications business proposition is significantly dissimilar in time division multiplexing (TDM) and Internet protocol networks, and the dissimilarity has profound implications for the capabilities required of OSS solutions. But all is not doom and gloom in the IP OSS tent.

Light at the End of the Tunnel?

Is there light at the end of the IP OSS tunnel? In practice there is rather more than a glimmer.

One of the first large-scale initiatives aimed at standardizing and integrating the OSS space was the International Telecommunication Union's Telecommunication Management Network project. This essentially postulated a four-layer functional model -- often depicted as a pyramid -- with, in ascending order, element management, network management, service management and business management domains.

There is some consensus that TMN was an important step in the right direction. However, most of its relevance is perceived to be in the bottom layers of the pyramid.

"TMN tried to put a lot of standards in between those layers. This never quite came to fruition, apart from right down at the element management layer," says Kosten Metrewell, London-based vice president of technical services at Orchestream, an IP network management and service activation company.

Also, the circumstance that each vendor could implement details of the TMN model in its own way, leaving service providers with the actual integration task, was not all that helpful. "Integration tended to be as big a job as if there were no standards in the first place," says Marconi's Kelly.

The TMF's Telecom Operations Map is a more recent initiative aimed at moving the focus of management toward business and customer management processes. "It serves as the industry example of how various processes are linked and integrated, and how they could be automated for flow-through," comments Jim Warner, president of the TeleManagement Forum, Morristown, New Jersey. The latest version, eTOM, adds things such as supply chain management and real-time customer interaction to extend the TOM framework to the embryonic commerce and e-business models.

TOM is itself the process definition piece of what the TMF terms new generation operations systems and software (NGOSS). This entity will marry generic business processes with off-the-shelf software, including extensible markup language (XML), and systems to produce intelligent automated, distributed end-to-end OSS capabilities, providing very rapid configuration, provisioning and delivery of new services. "It's not just plug-and-play flexibility but business operational flexibility," explains Keith Willetts, TMF chairman.

Like earlier OSSs, NGOSS will be required to perform differently for different folks.

For incumbents, NGOSS must leverage existing OSS investments but provide a clear path to the future, according to Wes Porter, vice president of telephony products for AP Engines Inc., a provisioning mediation solutions provider based in Maynard, Massachusetts. "NGOSS for the greenfield service provider must be very modular in nature, and must enable the service provider to make decisions to outsource parts of all of its OSS, but with the option to bring it in-house in the future," he says.

Unlike earlier OSSs, NGOSS will eschew protocols developed specifically for the telecommunications industry. "What we're looking at really for the next way of integrating is to use common industry standards -- open bus technologies such as common object request broker architecture (CORBA) or Java -- as part of the NGOSS," says Willetts.

Possible pain medications for OSS headaches ahead.

John Williamson (101741.2671@compuserve.com) is Global Telephony's Senior
Technology Editor in Chelmsford, England.

Coming Down the OSS Turnpike

Operations support system (OSS) industry players, watchers and customers recommend keeping an eye out for the following:

"The trends for service providers in IP QOS [quality of service] will be to limit the bandwidth consumed by customers in shared media networks, to offer tiered services with different pricing and bandwidth levels, to offer bandwidth management services, and to move to usage-based pricing." -- Azi Ronen, executive vice president for technology and marketing at Allot Communications, Tel Aviv, Israel.

"More integrated services. We're already seeing a little bit of a reaction against ‘best-of-breed’ OSS because it's such an expensive proposition and it's so hard to evolve over time." -- Tom Thistleton, executive vice president of corporate development and solutions delivery at Bluespring Software, a billing and OSS specialist in Cincinnati, Ohio.

"Application-specific SLAs [service-level agreements] as the only really meaningful SLAs. By creating an application infrastructure to sit on top of the network, a service provider can measure and track application performance and can enforce service-level guarantees so that users are
confident that they are being billed on a per-application basis and are receiving specific performance levels for each application." -- Todd Krautkremer, vice president of worldwide marketing at Packeteer, a bandwidth management solutions vendor in Cupertino, California.

"The ability to deliver QOS, including authentication and security, at very high speeds without slowing the network traffic down by making use of hardware-based IP switching." -- Martin Van Schooten, EMEA (Europe, Middle East and Africa) product marketing manager at Extreme Networks, a multilayer network switching solutions vendor in Santa Clara, California.

"According to my Harry Potter book of OSS, the two most important ingredients in the magic potion for effective NGOSS [new generation operations systems and software] are flexible mediation and an information model of the hybrid network." -- Avichai Levy, vice president of marketing at TTI Telecom International, Petach Tikva, Israel.

"More and more we see service providers deploying message bus architectures where they're not really connecting to any one application point-to-point, but rather communicating through a centralized message platform." -- Andrew Feinberg, president and CEO of Netcracker, Waltham, Massachusetts.

"You will have more sources of usage data rather than less. Increasing these other sources are going to be outside the control of the telco." -- Geoff Ibbett, business development director at Telecomms Consultancy & Solutions Ltd., Bracknell, England.

Cutting the Integration Tax

According to the TeleManagement Forum, the cost of integrating discrete operations support system (OSS) components is like a tax on the telecommunications industry -- it doesn't produce any value. Several approaches are designed to reduce that tax burden. Randy Custeau, vice president of products and technology at the Folsom, California-based OSI Division of Agilent Technologies, instances three.

The first is non-integrated, direct system-to-system connection between those "best-of-breed" business support systems (BSSs), OSSs and network elements that have relevance for each other. This, says Custeau, involves managing a large number of interfaces and synchronizing disparate data nodes, and is a maintenance nightmare because vendors release updates at different times.

A second approach is to insert an enterprise application integration (EAI) bus such as TIBCO or Vitria between the BSS and the OSS. This reduces the number of interfaces. But Custeau points out that you still need to understand what happens between the OSSs.

"What is the context? For example, what objects are they modifying? What functions are they performing? It's here where the difficulty lies," according to Custeau. In his judgment, this approach is good for BSS integration, poor for network integration.

A third approach, and one favored by Agilent, is to integrate the OSSs on an intelligent platform that includes mediation gateways and probes. In this scenario, the EAI becomes optional, the number of interfaces further decreases, and the deficiencies of the first two approaches are apparently remedied.

On the Upside

While there are undoubted difficulties attached to creating effective operations support system (OSS) solutions for Internet protocol networks, if these can be mastered, IP technology can lend itself to a much richer form of management. It's certainly more flexible.

In circuit-oriented systems, points out Jean Hammond, founder and chief strategy officer of Quarry Technologies, although the relative bandwidth available to different applications can be varied on a per-PVC (permanent virtual circuit) basis, these must be provisioned across the entire network, and there is no mechanism for setting relative priorities for a dynamic mix of applications and letting unused capacity be used for less important traffic.

"This is now possible with effective IP classification of traffic combined with highly granular policing and queuing," she says. "In many cases, these QOS [quality of service] attributes need to be enforced both in IP network cores and in local access by using COS [class of service] capabilities. The result is 'fast' provisioning of changes combined with the global reach of IP."

There's also an efficiency aspect. "The nature of IP, in terms of the information you can get out of the routers, basically allows you to monitor quality without having to use capacity to measure quality," says Dan DaCosta, Atlanta-based voice product line manager with Equant. By contrast, DaCosta says, in a time division multiplexing/signaling system 7 (TDM/SS7) network, to measure quality you need a box on the customer's premises and you have to generate calls on the network.

DaCosta also says the service-level agreement (SLA) on post dial delay is better on IP. "It takes less time to set up a call on IP than it does on TDM," he ventures.

Equant claims to operate the world's largest network, providing voice, data and Internet services in business centers in more than 220 countries.