June 1998

BY JEFF LAWRENCE

Service platforms designed to support the Intelligent Network (IN) and computer telephony integration (CTI) have been UNIX-based; however, recent developments have made service providers, carriers, and businesses more interested in using Windows NT-based platforms to support IN and CTI applications. Which platform is most appropriate in a given situation? That depends on several technical and business considerations. This article will review some of these considerations, along with some existing and future uses for various platforms.

FROM MONOLITHIC TO OPEN SYSTEMS
The telecommunications infrastructure is evolving toward an open, multi-vendor system. In the past, the intelligence was in the switch. Switch hardware and software tended to be monolithic, providing the switching fabric along with the intelligence to manage this fabric.

Service providers and telecommunications carriers provided services and applications. However, these providers found it difficult to introduce new services and applications because of regulatory, operational, and technical issues. And when these providers did introduce a new service, they found it difficult to move quickly. If and when a new service did become available, users lacked the ability to easily monitor and control it.

Today, carriers and service providers are opening up their networks in response to regulatory and competitive pressures. As Signaling System 7 (SS7) and Internet protocols (IP) become universally deployed throughout the network, new types of services, service control, and service monitoring will become possible. In the public telephony network, the IN is moving the intelligence out of the switches toward service platforms, and ultimately toward the users. CTI is providing applications on platforms that interact with users at their desktop via PCs and telephones. Internet platforms (in the form of servers) will provide a range of value-added services. It will not take long for IN, CTI, and Internet service platforms to be combined and provide a single platform from which a unified service architecture may be offered.

USES, APPLICATIONS, AND MARKETS
The demand for open platforms is being driven by the challenge of developing and deploying new applications and services quickly and cost-effectively. Service providers and carriers are providing local number portability, mobility, wireless, toll-free, Internet, and other enhanced services. Businesses are looking for ways to provide better service at a lower cost to customers.

In addition to IN services, CTI can help businesses toward their goals by offering additional capabilities for call routing, call queuing, call distribution, and productivity improvement. Open platforms can also be used as gatekeepers, gateways, and servers to enable Internet telephony, interworking between broadband and narrowband networks, telephony between IP networks and the public switched telephony network (PSTN), and mobile devices in different networks.

New service platforms will allow carriers, service providers, and ultimately the users to easily provision, monitor, and control services. Telecommunications Management Network (TMN), Simple Network Management Protocol (SNMP), and Common Object Request Brokered Architecture (CORBA) will be the major technologies used to accomplish this. Service platforms will also give users the ability to build Virtual Private Networks (VPNs) over the network infrastructure. The Internet model for VPNs uses the existing Internet protocols in conjunction with a server and some new security and tunneling protocols.

THE INTELLIGENT NETWORK AND CTI
The integration of Internet, IN, and CTI technologies will provide opportunities to offer a range of new services to businesses and homes. Standards and specifications are being defined by organizations, industries, and businesses to promote application and service portability, interoperability, and interworking. These efforts tend to focus on the call models, Application Programming Interfaces (APIs), and, to a lesser extent, bus architectures.

The CTI view of the world defines an overall architectural approach, call models, hardware models, and APIs such as Microsoft's Telephony API (TAPI), Sun's Java TAPI (JTAPI), the Enterprise Computing Telephony Forum's (ECTF) S.100 specification, and others.

The IN view is still developing. At the moment, there are some preliminary architectural models, provisioning models, call models, and APIs such as the Telecommunications Information Networking Architecture (TINA) specification or Sun's Java for Advanced Intelligent Network (JAIN) specification. There are differences between the various elements, and these differences, in conjunction with other technical and business issues, will influence the selection of particular open platform solutions.

OPEN PLATFORM ARCHITECTURE
The architecture for an open platform can best be described by the combination of host platform and operating system, APIs, internal bus architecture, and communications boards.

Host Platforms/Operating Systems
The host platform is typically a PC running Windows NT or a Unix workstation running a vendor-specific version of Unix. In both cases, Java Virtual Machines are also available.

APIs
There are several existing APIs for CTI and IN, including TAPI, JTAPI, and TINA. There are also efforts underway to specify a JavaBeans programming interface for the SS7 protocols used for the IN and wireless. Initial efforts have focused on the transaction capabilities application part (TCAP) and ISDN user part (ISUP) interfaces. Future efforts will likely focus on intelligent network application part (INAP) and message transfer part (MTP) Level 3. A JavaBeans interface for open platforms and the embedded environment will provide greater application portability and a future path for communications stacks to move into the Java Virtual Machine.

Bus Architectures
Typical bus architectures used for telecommunications equipment are VMEbus, PCI, and CompactPCI. These buses are favored for their bandwidth and existing or planned hot-swap capabilities. Other buses may be used to provide the data highways that interconnect the boards.

Communications Boards
Communications boards offer a wide range of microprocessor, digital signal processor (DSP), memory, and physical interface options. These boards typically run a commercial real-time operating system such as VxWorks, VRTX, and pSOS. Windows CE and embedded Java are also making inroads into the board market as their performance and real-time capabilities improve. In some cases, there are boards that eliminate the need for a host platform and simply operate as a single board computer running Windows NT or Unix.

PLATFORM CONSIDERATIONS
Service platforms for the IN traditionally have been Unix-based, and service platforms for CTI applications are usually Windows-based. This difference between the platforms used for IN and CTI, while it has come about because of historical and technical reasons, isn't the only consideration when it comes to platform selection.

The rapid changes taking place in the network, along with the businesses and organizations using the network, are influencing the nature of service platforms. Specifically, there is an increasing interest in Windows NT platforms for IN and Internet services.

We review several platform selection considerations in the subsections below. As we will see, these considerations can be addressed through a variety of architectural approaches. In general, five different types of solutions that are offered for open IN platforms

Features
For the most part, the features supported by Unix and Windows NT platforms (subject to hardware capabilities) are the same. Within a platform, however, there are still issues of scalability, port count, physical interface support, multiple protocol support, conformance to standards, and interoperability with other products.

Performance
Service platforms have high performance requirements. Unix and Windows NT have not traditionally been considered as strong choices for real-time applications. Unix and Windows NT performance varies depending on whether the application or communications stack resides in the user or kernel space. It also varies depending on how the functionality of the applications and communications protocols are distributed between the host and interface boards. The lower layer protocols of SS7 are processor intensive and typically are best executed on an intelligent board running a real-time operating system. Usually, performance decreases, ease of development increases, and overall system reliability increases as the communications stack moves closer to user space.

Reliability And Availability
Service platforms in the public network have very high availability requirements, whereas platforms within enterprise networks usually do not require such high availability. Unix systems have offered several hardware and software solutions to achieve high availability. Windows NT machines have been perceived as being incapable of achieving the same level of availability, although recent advances in clustering and other software architectures are addressing this issue. Hardware solutions for high availability are typically more expensive than software solutions.

Quality
Quality is dependent on the underlying hardware and software providers and the successful integration of their technologies.

Security Of Supply
There are many different flavors of Unix, and easy movement of applications among them is not guaranteed. The different Unix platforms tend to be manufacturer-specific and tied to a particular hardware solution. Windows NT is not tied to a particular hardware solution. This provides a higher degree of flexibility and negotiating leverage with hardware vendors.

Most SS7 solutions are provided as binary code solutions. Source code solutions potentially remove any dependence on the technology provider, provide the ability to support the software directly, allow the platform user to add their own features, and potentially allow migration of the application and communications stack into other communications equipment.

Ease Of Application Development
The speed and ease of application development is very dependent on available tools, supporting applications (such as databases), and application development environments. This can differ significantly depending on the underlying operating system. Another factor that can ease application development is the availability of software engineering knowledge and skills for the tools, supporting applications, and application development environments. This can be very important in a resource-constrained world.

Flexibility And Application Portability
An important consideration in today's open environment is the ability to move and scale applications and services between different platforms. Application portability and reuse is dependent on the APIs to which the application is written, the underlying hardware buses used for communication between boards, and the underlying microprocessor technology.

Lifecycle Cost
There are several aspects to lifecycle cost, including development, deployment, and support costs. Development costs are typically dependent on the ease of application development. Deployment costs are dependent on the underlying hardware and required availability. PC-based platforms are typically lower cost, although high availability will increase the cost. Support costs are dependent on whether the solution is binary or source, and whether the platform is PC or workstation/server based.

Future Proofing
Future improvements in tools, development environments, and performance are dependent on the commitment of the underlying operating system supplier. Future improvements in performance are also dependent on the underlying microprocessor technology and the commitment of the chip manufacturer to improving that technology.

CONCLUSION
There is a place in the network for Unix and Windows NT open platforms. These platforms are needed to support IN, CTI, and Internet applications. As these applications proliferate, open platforms will become all the more necessary.

While the need for open platforms is clear, it is not so clear which open platform is best for a particular application or service. Often, the selection of an open -platform solution presents a trade-off.

For example, binary code solutions typically have a lower up-front cost. However, the lifecycle cost of binary solutions may be greater than source code solutions in the long term. Binary code solutions are typically more turnkey and will take less time to get to market. Source code solutions will require some integration work and take more time to get to market, but ensure far less reliance on the technology provider and greater flexibility for development of new features and maintenance support.

Which open platform you use reflects both your technical requirements and your business priorities, particularly as they relate to short-term and long-term needs. Balancing these needs, and selecting the most appropriate platform, can be a challenge. You will, for example, need to keep time-to-market pressures in perspective.

Carriers, service providers, and equipment manufacturers are all feeling strong pressure to quickly and efficiently develop and bring their products to market. However, other considerations in addition to the time-to-market imperative may also weigh heavily. Each application or service requiring an open platform will present a unique set of considerations. Weighing them all appropriately, and selecting the most appropriate open platform can help ensure the success of carriers, service providers, and businesses.

Jeff Lawrence is president and CEO of Trillium Digital Systems, Inc., a leading provider of communications software solutions for computer and communications equipment manufacturers. Trillium develops, licenses, and supports standards-based communications software solutions for SS7, ATM, ISDN, frame relay, V5, IP, and X.25/X.75 technologies. For more information, visit the company's Web site at www.trillium.com.