October 1999

TAPI, Take 3: Third Time's The Charm

BY JIM MACHI

[Go To Sidebar: What Is TAPI 3.0?]

The soon-to-be-released third major version of Microsoft’s Telephony API (TAPI) represents a coming of age for this often misunderstood technology. It is a critical component of an all-out attempt by Microsoft to reposition its family of operating systems toward high-end, mission-critical CTI solutions.

A SOLUTION SEARCHING FOR A PROBLEM?
As with many other technologies, early incarnations of TAPI were often criticized as a “solution in search of a problem.” The concept was great: Provide the necessary plumbing for disparate telecom applications to control heterogeneous telecom peripherals. But the implementation was not powerful enough to really be useful.

Although TAPI successfully leveraged Microsoft’s Windows Open Services Architecture (WOSA) and abstracted applications from the underlying hardware and telecommunications protocols, it was plagued by the instabilities of early versions of Windows and Windows NT. Also, in an effort to secure widespread support, TAPI was designed as a superset of the way all the major telecommunications manufacturers implemented features such as transfer, conference, and hold. This made it extremely hard to implement true heterogeneous solutions, since each manufacturer could still implement things in a pseudo-proprietary manner. Plus, the API was limited to C function calls and was difficult to map into today’s object-oriented methodology, or to be used by higher-level programming tools such as Java or Visual Basic.

Granted, TAPI 1 and TAPI 2 did find their way into some smaller call center and interactive voice response (IVR) applications. These earlier versions of TAPI were particularly good at first-party call control from the desktop, and helped launch the short-lived explosion of “soft phones.” However, TAPI was still very underpowered and somewhat immature to be considered for high-end, industrial-strength, server-based telecom solutions.

Even with TAPI’s shortcomings, it was clear from the beginning that Microsoft was on the right track. Few doubted Microsoft would some day make good on its claim to evolve NT so that it could be taken seriously in the telecom arena formerly reserved for UNIX and proprietary operating systems. TAPI is a major piece of that strategy.

READY FOR THE BIG LEAGUES
TAPI 3.0 (along with other communication technology initiatives and their overall focus on improved reliability, availability, and scalability), put Microsoft in a position to make a full-scale assault on the telephony establishment in Windows 2000. Microsoft is also making a focused effort to integrate as much IP telephony functionality as possible into version 3.0 of TAPI. IP telephony is poised for explosive growth as organizations begin the historic shift from expensive and inflexible circuit-switched public telephone networks to intelligent, flexible, and inexpensive IP networks.

From its humble beginnings in the Intel Architecture Labs, and its release by Microsoft in 1993, TAPI has grown steadily with each successive release. The first release, version 1.3, provided basic first-party call control. It was deployed as an add-on to Windows 3.1 and NT 3.51. This version adequately achieved its objective of abstracting applications from devices and network protocols. It was followed in 1997 by version 2.1, which was released as part of NT 4.0, and found a niche in solutions requiring higher-level, adjunct-based call processing applications such as medium-size call centers and enhanced services. Version 2.1 expanded on TAPI’s fledgling third-party call control model and added considerable client/server and automated call distribution (ACD) capabilities. These enhancements were needed to implement sophisticated call center solutions and perform the elaborate call manipulations needed for non-traditional billing schemes such as prepaid calling, credit card calling, and pay phones. Besides providing an easier way to centrally manage and report on the system’s performance, these new capabilities successfully deepened the abstraction for applications from the proprietary nature of PBXs and other legacy telephony systems.

As convergence begins to gather steam, TAPI is poised to take yet another evolutionary step. It is now an alluring component for use in large-scale, mission-critical solutions integrating heterogeneous and disparate video, data, and voice networks. The ability of TAPI 3.0 to direct conversations over packet protocols adds a powerful abstraction for applications, essentially freeing them from the details of the underlying media being manipulated. The quality of service (QoS) mechanisms provided as part of TAPI 3.0 will also create unique and more robust applications.

It is this latest abstraction that will catapult TAPI into the mainstream of CTI development. Although the idea of open systems, pioneered by the computer industry, was long shunned by the telecom establishment, it is now sweeping like wildfire through the industry. As the world continues to transition from closed, proprietary telephony platforms to open, standards-based CT servers, developers and system owners are looking to protect their investments. Thus, hardware and application abstraction becomes a fundamental requirement for telephony solutions. TAPI’s newfound media abstraction capabilities help isolate applications from changes in the underlying network and switch fabric. As technologies like ATM, xDSL, and even voice or fax over IP begin to permeate the network, system owners can preserve their investment in applications and developers can be assured that much of their work will be protected.

WHY TAPI?
There are several market and industry trends contributing to the increasing importance of TAPI relative to other CTI abstraction mechanisms. These include:

• Win32 familiarity. TAPI is based on the universally-known Microsoft Win32 API. Therefore, it is immediately familiar to — and easily learned by — the 3.2 million professional Win32 developers in the industry today.

• Platform nativization. The Win-dows operating system is constantly incorporating new development tools and underlying services that simplify the process of creating applications. This makes TAPI, as the native telephony API in Windows, more desirable to developers. As Microsoft embraces more of the underlying functionality in its OS and services, developers can continue to back out of proprietary code, lowering their support liability and focusing instead on the higher-level functions of their application — where they add the most value.

• Support from Microsoft. TAPI ships free with every copy of Windows NT, and has Microsoft standing behind it — unlike any other major telephony programming environment.

• Ever-increasing portfolio of TAPI-compliant hardware and software. Dozens of companies have already pledged support for TAPI, and many are already shipping TAPI-compliant products. Microsoft has even developed a separate logo program for TAPI. In addition to the current support for TAPI by companies like Intel (formerly Dialogic), Brooktrout, AltiGen, and Natural MicroSystems, Microsoft itself is driving demand for TAPI by delivering platforms and applications built around its TAPI technology. These include the Web Telephony Engine service in the upcoming Windows Platform Software Development Kit for Windows 2000, and the next version of Windows NetMeeting.

NOW WHAT?
If you are new to CTI, or actively working on an IP telephony solution, you should consider TAPI. If you are an experienced CTI developer and once flirted with TAPI, but found it not up to the task, you should take another look at TAPI 3.0. It is very likely you will find that TAPI has come of age and is now ready to play in the majors.

Jim Machi is director of product marketing, Internet Telephony, for Dialogic Corporation (now part of Intel). Dialogic is a leading manufacturer of high-performance, standards-based computer telephony components. Dialogic products are used in fax, data, voice recognition, speech synthesis, and call center management CT applications. The company is headquartered in Parsippany, New Jersey, with regional headquarters in Tokyo and Brussels, and sales offices worldwide. For more information, visit the Dialogic Web site at www.dialogic.com.

TAPI 3.0, which has been in beta testing for almost two years, will ship with Microsoft’s new operating system, Windows 2000. Client support will be made available on the Windows 2000 family of operating systems, along with Win-dows 98. Since TAPI is an evolutionary API, Microsoft has guaranteed backward compatibility with applications and hardware written for the older version of the specification. TAPI 3.0 adds four major new features that most developers will no doubt want to embrace:

1. COM interface for language-independent support (in other words, it can be used by VB, Access, Borland Delphi, and Java, as well as by C/C++ developers).
2. Media streaming support for access to computer telephony media.
3. Complete IP telephony support. TAPI 3.0 will ship with an H.323 stack and IP multicast that will work with NDIS-compliant network interfaces, to enable IP telephony applications to be written faster, more easily, and less expensively. The T.120 protocol for data sharing and the G.711 codec will also be included.
4. Tight integration with other Windows 2000 features such as active directory service to simplify deployment within an organization, and QoS support to improve conference quality and network manageability.

For additional information on TAPI 3.0, please refer to the Microsoft article, Tapping Windows 2000 for TAPI 3.0.