By: David J. Wardell

© 1992 By: David J. Wardell.  Reproduction or redistribution in any form without written permission is strictly prohibited.

Retrospective

A number of years ago I wrote several pieces discussing Sabre's TravelBase travel agency accounting and MIS platform.  At the time the material was well received and considered insightful.  In later years, however, a few critics developed selective memories and criticized me for "endorsing" TravelBase when (at least in their minds) the system was a failure.

Here is what I said about TravelBase at the time.  The key themes, as expressed near the end of the piece, are these:

1. TravelBase had yet to be completed, let alone deployed at the time (1992).  Clearly everything depended upon how successfully the vendor translated concepts into reality.  In practice, execution sadly lagged development and concept.  Moreover, nine years (as of this writing) is an eternity in data processing and the notions that seemed innovative at the time appear lame by today's standards.
2. Successfully using TravelBase required commitments by users that were new and much more demanding than had previously been required.  With characteristic lack of planning and commitment, most users doomed themselves to failure by their own ill-conceived execution.
3. Real-world conditions were far different than had been assumed through development of the system, hence performance and usability as conceived did not play well outside the laboratory.

It's happened before and will again: poorly planned and executed technology projects show considerable promise in the development stage, but collapse once they are deployed.

Overview

In early June (1992), the Sabre Travel Information Network (STIN) announced Travel-Base--the most important travel industry software product in many years.

TravelBase is the successor to STIN’s highly successful ADS (Agency Data Systems) travel agency accounting product and represents an effort to introduce current-generation hardware and software into the world of travel financial controls and MIS.

The travel accounting market has been saturated with products for several years--so much so that no innovative new products have been introduced for a long time.

ADS has dominated that market almost since its inception because it is clearly the most successful travel data management package--meaning the combination of hardware, software, support and enhancement (this definition is important)--ever produced.

Competing systems used interesting approaches but failed to deliver a comprehensive package that allowed agents to tap their features and continue using them successfully.

Based on technology concepts from the mid-1970s, ADS was outpaced by developments in the computer industry.

Further, today’s travel information and business requirements are much more demanding than they were when ADS was new.

TravelBase intends to take advantage of the best in current hardware and software tools, while preserving the important parts of the support package that made ADS successful.

In a departure from past practices, it will incorporate components from many sources and not rely exclusively upon internal development. It is, in this respect, the first genuine attempt to produce an "open" travel information product.

Design Goals & Concepts

STIN’s design goals for TravelBase are:

1. Industry standards and connectivity should be supported throughout the project.
2. No proprietary products should be used.
3. A multitasking environment should be supported.
4. Full interoperability with other STIN products (notably Sabre) should be supported. Interoperability means that a program (application) can be run on various hardware and software configurations.
5. Network-based batch processing and task scheduling should be supported.

The adoption of true industry standards and non-proprietary products (at least as far as the basic platform is concerned) is certainly appropriate and will add to the system’s long-term success and sustainability.

By now you should understand that TravelBase is a major departure from "business as usual" travel software projects.

It is using "leading-edge" designs to produce a product with new capabilities not previously offered to the industry.

TravelBase uses the popular client-server approach to data base management.

In substance, a data base server is responsible for managing data base requests quickly and consistently.

The server approach used by TravelBase consists of software that controls access and updates to the database.

Applications running on separate hardware (workstations) are connected to the server using a LAN (local area network) and make "requests" to the server, based on the needs of the applications that run on the workstations.

Programs requesting services from a server (in this case access to a data base) are called the clients of the server; there are no "clients" unless there is a "server."

When used to control databases, the client-server approach allows the server to manage sessions with multiple clients, meaning that many workstations doing different tasks can access the same physical database efficiently.

The client makes requests to the server using the native data base language, such as SQL (structured query language), as used in TravelBase. Users run programs on workstations and communicate the results of their work over the LAN to the central database, controlled by the server.

The advantage of this approach is that development of attractive user applications is greatly simplified and can be customized to meet the needs of specialized tasks, while the central data files are standardized.

All users have access to the same data, regardless of how applications running on their workstations appear. The agency also benefits from better data management, computer efficiency and error control.

STIN has selected Sybase as the data base development and server-based support environment. It is a highly successful commercial data base product that implements a number of advanced features.

Symmetrical Multiprocessing

Among these is the ability to take advantage of SMP or symmetrical multiprocessing. Multiprocessing describes a system capable of managing multiple processors. (It does not mean running multiple processes, which is multiprogramming.)

Multiprocessing solves one of the oldest problems with the basic computer architecture: the von Neumann bottleneck, which results when a computer funnels all activity through its central processor.

A multiprocessor system allows several processors to coordinate their activities while maximizing system performance and throughput.

By contrast, multitasking allows the computer to handle several applications simultaneously while the programs share a single processor.

Without getting into a detailed discussion of SMP, it is important to understand that SMP is but one approach to multiprocessing. In a SMP system, all processors are peers with equal access to system resources and memory.

Most modern software designers believe that SMP is a correct and effective approach for achieving high system performance at a reasonable development and operational cost.

UNIX

TravelBase uses UNIX as the operating system for its server. This is consistent with the system’s orientation toward non-proprietary products ("open systems")--most people really mean UNIX when they think of "open systems."

I’ve personally advocated UNIX for many years; its implementation in a major new product is appropriate and welcome.

The only point of criticism I could offer about the TravelBase operating environment is that there is not more UNIX used--TravelBase will employ OS/2 as its LAN-based workstation operating system instead of implementing UNIX throughout its design.

TravelBase envisions a symmetric multiprocessing environment, running on a high-end UNIX machine.

Hardware

Data General (manufacturer of most prior ADS hardware) is the announced vendor of choice.

TravelBase will use DG’s new Aviion hardware products. These systems differ substantially from the older MV or Eclipse systems so familiar to agents.

Data General-left much of its proprietary development orientation behind as it embraced "open systems" (meaning UNIX) and implemented new technology.

It is important to note the distinctions between Aviion products and DG’s older system: While agency-users are blessed with much more powerful and apparently more reliable hardware, these advantages will be realized in Travel-Base through systems that are significantly more complex than agents are used to and that will require a different skill base to be used successfully.

This is a very important implementation issue.

Using UNIX-based Sybase as a powerful relational data base management system and on-line transaction processing system, TravelBase also will use IBM’s OS/2 operating system.

Workstations & OS/2

Individual workstations actually will run the OS/2-based TravelBase applications; the UNIX-based data base system will be accessed as a server, where necessary.

This "distributed processing" approach, if properly implemented, should create a flexible and powerful work environment for agency users.

I should spend a minute evaluating the OS/2 decision and here some background is in order.

In 1987, IBM, in partnership with Microsoft at the time, designated OS/2 as the PC-based operating system of the future that would eventually replace MS/DOS.

This imperfect early release was modified considerably when version 2.0 of OS/2 appeared earlier this year.

The "new" OS/2 takes advantage of present user interface concepts and current PC technology.

Microsoft and IBM subsequently chose to develop competing "new" operating systems for the PC, with the former choosing instead to pursue enhancements to its Windows environment.

The most advanced Windows technology now under development, termed Windows NT for "new technology," incorporates many of the most important features that also are available in OS/2 and several that are not.

Windows, and the applications that use it, have been substantially more popular than OS/2 or its applications.

The end of all of this is that a robust future for OS/2, both in terms of future enhancements and available third-party applications, is by no means assured.

I have never been a fan of OS/2, particularly in highly specialized, production-based applications.

The path to realizing the fullest benefits of leading edge hardware and software has always been shorter, clearer and more sustainable through UNIX.

While I can appreciate why OS/2 was selected for TravelBase, I would personally have been more pleased with UNIX-based applications on both the server and the workstations.

It is true that there probably will be more third-party software development in the travel industry that is compatible with OS/2, but this benefit is largely illusionary: The actual number of users that will run third-party programs of whatever description on their production accounting system network is fairly small.

Accounting & Databases

TravelBase applications generally encompass the full range of travel accounting and data management tasks represented in modern accounting and MIS systems.

The relational structure of the database will make data more accessible and extend reporting flexibility that has previously been unavailable in most travel accounting products. STIN has elected to develop most of the actual financial and reporting applications in house, instead of adapting existing products.

Whether you agree with this approach or not, its justification rests upon the ultimate success of the applications, which must await formal release of the system.

Preliminary information on planned applications indicates that the initial phases of the product will cover the accounting and MIS basics, albeit in a more efficient manner, while future plans promise expanded functionality and features.

User Interface Issues

TravelBase will make extensive use of graphic interfaces, similar to the technology used in Windows or Macintosh-based programs throughout its applications.

If you have studied this subject at all, you know that there are conflicting views as to the overall efficiency of graphic interfaces for production systems.

STIN is aware of this and consciously elected to pursue the graphic course.

Overall Impressions

As a result of several columns I've written, a number of people have asked, "It is (TravelBase) really that good?"

The answer is, assuming that the eventual release of the product is successful, TravelBase will significantly expand travel accounting and MIS capabilities beyond those available from most competing products, and will, in fact, render many such products hopelessly obsolete.

To speak frankly, however, it is not the ultimate in DOS-based travel management, as I have envisioned it frequently in these pages.

In evaluating TravelBase, there are a number of specific concerns that affect the usability of the system for any individual agency:

• Expertise. The system’s design requires extensive staff training and reorientation, which will not be achieved without cost and effort.
• It will also require a different skill-set for the STIN organization to support.
• Conversion. Switching to TravelBase from ADS, or competing products, will require commitment, effort and planning. It will not happen automatically.
• Performance. While initial projections are promising, delivery of the first systems must show how well the hardware and software perform under real-world conditions.
• Execution. This industry has seen well conceived and well-intentioned projects come up short or be abandoned. Only when TravelBase, as promised, is shipping will its concepts be truly proven.

TravelBase should begin installing by the end of the year. It represents an innovative and progressive approach to travel accounting and MIS, certainly one worthy of your consideration, as part of an overall technology strategy.