Many more are actively evaluating NEXTSTEP, while others are in the process of building object libraries and assembling their applications. This is a useful time to take a close look at the way early corporate adopters are going about the process of implementing NEXTSTEP.
First, consider the reasons that information-systems strategists are receptive to NeXT's message. In the 1990s, businesses are looking to information technology, especially software, to change the competitive alignment in their industries. Most businesses are also at a transition point in the life of their computer systems. They are moving from the disjointed world of back-office mainframes and desktop personal computers to the promised land of distributed client-server computing.
As a result, everything in software – operating systems, databases, development tools, applications – is up for grabs. In the operating-system arena, the basic requirement is a 32-bit multitasking system, whether it be Windows NT, OS/2, or any flavor of UNIX.
NEXTSTEP fits well with the client-server model. It runs on industry-standard hardware platforms and offers the unique advantages of the best object-oriented development tools available. For companies in competitive industries that are facing systems transitions anyway, considering NEXTSTEP is no longer a far-out alternative but rather a straightforward business case.
That gets NeXT in the door. Ramping up to full-scale deployment can take as long as three years, with decision points at several stops along the way. First, companies bring in a few systems for evaluation and prototyping. If all goes well, the company may commit to a large purchase, but the vast majority of seats are not actually purchased until the successful completion of the development phase.
From NeXT's point of view, the sales cycle is a straight correlation between time and number of seats. A sale of 50 seats might take three months, but a corporate rollout of 1000 seats takes two years or even longer. NeXT cites three sales case studies (see chart) that show an average of about 30 months from first contact to deployments in the high hundreds.
To get a sense of the experiences customers are having as they move through the NEXTSTEP cycle, NeXTWORLD spoke to managers at a selection of major NeXT customer
sites. We also prepared our own unscientific time line, a composite of the experiences of the customers we interviewed for this article.
It used to be axiomatic that no information-systems manager was ever fired for choosing IBM. But the days of playing it safe in technology are long past. Today, companies look for computer systems that give them a competitive advantage. Failure to take a risk may be the biggest risk of all.
Nowhere is this more true than in the market for mission-critical custom applications, where NeXT does battle for the hearts and minds of corporate-technology strategists. And more often than not, when NeXT is invited to compete in an evaluation process for a strategic new application, the company is finding a receptive audience.
Still, the decision to take a risk on NeXT is one that requires equal measures of vision and intestinal fortitude. Customers who take the plunge know that they are out on the bleeding edge of technology. As a result, the process of evaluating, testing, and committing to NEXTSTEP has typically been a long, drawn-out, and sometimes agonizing cycle that lasts from several months to a few years.
Tough questions According to NeXT, these are the questions most frequently asked by potential customers who are evaluating object-oriented technology and NEXT-STEP:
How do you find the answers to these questions? For starters, says Warren Weiss, NeXT's vice-president of North American sales and marketing, talk to other clients. "My single biggest recommendation is to call customers early on, so you don't have to rely solely on the word of salespeople."
Secondly, recognize that changing to a new technology will require partnerships with VARs and system integrators. While most companies already have established partnerships, NeXT is in the process of expanding its roster of support providers, which it will recommend to clients looking for collaborators, Weiss says.
In addition, customers should understand that there will be pain involved in implementing NEXTSTEP or, for that matter, any new technology. One good measure of how much pain there will be and whether it will benefit the company's bottom line is actually building a NEXT-STEP application. Customers who are serious about the technology have been building prototypes using NEXTSTEP and other development environments under consideration as a way to compare the various technologies. Company executives can then see if the applications they envision can really be created using a particular technology. "Customers have generally explored every traditional method and every other format – from COBOL to C++ – prior to getting NeXT involved," says Weiss.
"Because they're usually involved in high-risk projects that will affect the bottom line, they want to see a prototype. But everyone else throws the prototype away because they're just screens. Our prototype, because it's built using true object technology, is the first good iteration of the project."
NeXT won't build the prototype for customers, but, Weiss says, "we look over their shoulder." Potential customers can participate in NeXT's mentorship program, a ten-week developer camp in which they can learn to use object technology and get help creating the specific programs and customized apps they desire.
Banking on NEXTSTEP In the case of NationsBanc-CRT, the securities-trading arm of the Virginia-based banking giant, the prototype was one of the final steps in a nearly year-long evaluation process that led the company to make a long-term commitment to NEXTSTEP in May, according to John Keazirian, executive vice-president of information technology.
The process began with the company identifying the goals and benefits it hoped to achieve using object-oriented technology for applications that involve pricing securities, risk management, real-time information distribution, and operations and settlement processing. In order to remain competitive, NationsBanc-CRT recognized that these complex applications needed to be developed faster and would have to run on multiple platforms that would put more computational power on traders' desks. This interoperable environment, filled with custom applications, would be part of a real-time, global network fueled in part by high-powered servers.
After sending many of its in-house staff through NeXT's developer boot camp, NationsBanc-CRT decided that NEXTSTEP was superior to the other technologies under evaluation. "NEXTSTEP was a mature, top-to-bottom object-oriented technology. There weren't any other ones out there, with the exception of Smalltalk. But Smalltalk doesn't fit in with our environment because of the performance penalties you pay on the calculation side," says John Bruns, vice-president of technology. "We tried a number of builder technologies to create the prototype and there was no comparison [to NEXTSTEP].
"Developers felt they were able to take the project and rework it a couple of times in the same amount of time they thought they were going to build an application. As a result, we already have a much more robust object library, and the prototype is pretty close to being the final application," he says.
Business considerations But NEXTSTEP did not win its place at NationsBanc-CRT on its technical merits alone. "We had a number of other criteria we were looking at," adds Keazirian. "In terms of their software and hardware, we were concerned that they would be a proprietary solution. Would they have the resources to drive the hardware and give us the kind of power we need? What was NeXT's viability as a firm?"
NeXT's Object•Enterprise alliance with Hewlett-Packard solidified the deal. The company also approved of NeXT's decision to abandon its hardware business and focus exclusively on software.
"We felt they needed to concentrate on being a software vendor, on opening up new systems and increasing interoperability, and having an alliance with HP, because we felt that HP provided the best scalable solution with a RISC architecture," Keazirian says. "We were in close contact with NeXT as it was going through its transition and, point after point, we saw them falling into a business plan that we felt was important for them to survive."
Near hits But even though NEXTSTEP has repeatedly scored the highest marks, technologically speaking, during the evaluation phase, it has failed to make the final cut at several companies for a variety of reasons.
"We looked around and tried to figure out who was a leader in the field as part of an investigation to see if object technology was real in a business environment," said one executive at a major financial institution in New York.
"We started work with NEXTSTEP and 75 percent of what NeXT told us was true: It was easy to develop, there was seamless integration, and there were resources to support us.
We set out to accomplish something: Deliver an application on time to one of our customers and have it implemented. We were able to do that. But when you start getting into the next level of investment, that is, whether to make it a corporatewide technology on our trading floor, that's where other considerations come into play – politics.
"We are still using object-oriented technology and we consider our NeXT experience extremely valuable. But we've stopped using NeXT because we require flexibility without worrying about which platform will succeed. It's easier to buy hardware than develop software, so the fact that NEXTSTEP doesn't run on Macs and Suns was an issue."
NeXT knows it won't win every sale. For every company that is willing to take a controlled risk in exchange for huge potential benefits, there are many more that seek the comfort of the safe choice. Today, the safe choice – the one you won't be fired for making – is Microsoft.
On the other hand, most information-systems managers today understand that the big payoff from technology is a strategic advantage. After going through a careful evaluation process of gathering information, comparing technologies, developing a prototype, and lining up third-party support, an increasing number of far-sighted companies are choosing NEXT-STEP, and not looking back.
Connie Guglielmo is a free-lance writer and editor.
If some customers fall in love with NEXTSTEP during the evaluation stage, then the development phase is like coming home after the honeymoon: Now you live with it. Warren Weiss, NeXT's vice-president of North American sales and marketing, says that decreasing the time to market for a product is the overriding concern for most NEXTSTEP customers. And the development stage – in which the down-and-dirty coding happens – is when the NEXT-STEP time savings are supposed to start accruing.
The good news is that most customers do report exactly what NEXTSTEP salespeople have promised: faster development cycles.
Faster, that is, once the in-house staff is up to speed on object-oriented design and NEXTSTEP programming. Faster, that is, once the staff's needs assessment has been recalculated in terms of a new development environment with new capabilities. Faster, that is, except for the most time-consuming elements universal to application development: staff training, needs analysis, architecture design, and user testing.
Choose your partner Since such requirements vary from site to site, there's no such thing as a "typical" NEXTSTEP product life cycle, especially during development. But the vast majority of NEXTSTEP customers are object-oriented-technology neophytes and need to build a lot of additional training time into the implementation schedule, often by bringing in outside partners such as systems integrators, VARs, and consultants, or all three.
For customers who really need to meet NeXT's estimate of three to six months for developing a custom application, the quickest method is simply signing on hired guns to provide a finished product. That's what PanCanadian Petroleum did for its Gas Production Inquiry and Reporting System, when it decided that the application had to be completed in five months.
Since its relationship with the team of consultants has been so critical to the final outcome (four in-house projects are due to be deployed in April), Jocelyn Barr, PanCanadian's coordinator of applications architecture, emphasizes the importance of choosing outside partners early in the game.
"We ought to be there, if not day one, then day two or three," agrees Frank King, president of Austin, Texas-based Pencom, a system integration house that has handled nearly 30 NEXTSTEP customers.
To increase the number of partners available to custom-development customers, NeXT recently hired five Object Channel representatives who now devote their time to recruiting and assisting system integrators, VARs, and VADs.
Slips happen According to interviews with customers and system integrators like King, working with such an outside partner can help a company avoid the three major pitfalls to organizing a NEXTSTEP project:
When the NEXTSTEP license was signed, William Morris was not only new to NEXTSTEP, it was new to computers, period. Working with outside consultants under his supervision, Henry has directed the development of several integrated modules that automate almost all of the agency's everyday functions, from tracking the 200–300 phone calls a talent agent makes in a day to matching actors' résumés with cast descriptions.
In the case of William Morris, the mistake was "overengineering" the first version of the interface, according to Henry. The second rev of the first two modules, which includes simplified interfaces, is now being beta-tested and will be deployed this fall. Five more modules are scheduled to be deployed this winter.
For these reasons, most companies discover that it really pays to allot sufficient lead time for designing the architecture and overall system infra-structure. For first-time NEXTSTEP customers, this means both a more flexible schedule and allowing for some organizational changes during the development phase.
"Customers want the expertise in-house at the end of the day. It turns out that mentoring is a much bigger concern for people than we would have thought a year ago," says Pencom's King. He reports that, instead of straight programming consulting, most of his customers now want joint development teams organized to transfer his programmers' NEXTSTEP knowledge to in-house development teams.
Proof of concept NeXT itself recently unveiled a mentorship program that provides a 10–14 week "education" designed to teach NEXTSTEP by helping in-house staff prototype the company's custom applications, instead of using generic training exercises.This method also provides customers with the chance to "throw away" their first design attempt.
"If everyone got rid of their first application, they'd be better programmers," says Scott Weiner, mentorship program manager. Most users claim that the extra time for all this training and experimenting is partially recovered by the technology itself. With NEXTSTEP's Interface- Builder, interface design and other coding tasks are indeed faster than using conventional programming languages. And even in cases in which a company's first effort isn't record-breaking in terms of productivity, time is saved in maintenance reduction or reusing generic objects to build the next application.
Schedule gains for a single project can be made as well by designing prototypes not just as proof-of-concept tools, but as building blocks for the finished system. NeXT executives stress that prototyping under NEXT-STEP should be considered part of the development phase, not a final evaluation test.
"Everyone else takes the prototype and throws it away. [NEXTSTEP] prototypes close the window of the product life cycle," says Weiss.
NEXTSTEP customers agree that NEXTSTEP prototypes are time-savers. Even if the initial prototype is just an interface shell, that shell can later be tacked on to the back end without major modification. Pencom's system integrators design applications as a cycle or "spiral" of prototypes, with each level building on and refining the last version, according to King.
Mt. Clemens Hospital used a similar methodology. First, it designed the data tables intended for use, then had end users populate the tables at the same time the basic architecture was being designed. "That way we didn't waste time," says Chowdhry.
Chowdhry's team, however, did end up needing a lot of additional time to build in three levels of redundancy for the CPU, ensuring a system-crash recovery time of 45 minutes. Since the hospital runs on this system, its reliability is essential, even if ensuring it prolonged the development cycle.
Indeed, most custom-development efforts call for something uniquely essential to that site. Such requirements are almost invariably time-consuming, but the customer ends up with a better product.
"The productivity gain that NEXTSTEP gives you can be used in two ways: to do a particular task more quickly or to do a more sophisticated task. All of my clients have opted for the latter," says King.
Clair Whitmer is the West Coast correspondent for the IDG News Service.
For all the concentration on software and hardware that goes into the early stages of the NEXTSTEP cycle, the most important consideration in deploying any new technology is people. Whether you are writing custom apps, building new networks or upgrading old ones, or rearchitecting your entire business, technological decisions should not be divorced from the needs of users.
Start early One of the easiest ways to ensure a successful rollout of a NEXT-STEP project is to involve as many end users as you practically can during the evaluation and development stages. "We have rarely had problems with deployment because we've worked closely with our traders, who are our users. We live on the trading floor," says Hadar Pedhazur, managing director of global equity derivatives for Union Bank of Switzerland (UBS) in New York.
Problems still arise, Pedhazur notes, from deploying a custom system to offices in different parts of the world, but an agreed-upon common look and feel is the essential starting point. Differences are hammered out in con-ference calls and occasional meetings of all the involved players to reach a "peer-group agreement," he says. It is far easier to tweak an application for differing legal practices in various countries, for example, if the template for factoring these practices into the application is chosen early on.
Dwight Koop, executive director of information technology at Swiss Bank Corporation (SBC), agrees: "It's a simple metaphor. If you use a futures trading app in Chicago and go to trading stocks in London, the app changes but the look and feel is the same." Koop explains that this is only a logical extension of the reusable-code capabilities of NEXTSTEP, because "forcing creative people to reuse other people's creativity is okay if they can work faster."
And users who are consulted early will understand why new software doesn't work the way it should, explains Mark Richards, president of Affinity Software, a spin-off of Alain Pinel Realtors in northern California. "We thrust a lot of betaware on our users. They were gracious putting up with bugs, and continue to do so because we're always putting up new things," he says.
Technology of champions Clearly, winning the support of end users comes from more than simple involvement. Commitment from corporate heads and team cohesion both contribute to, and result from, user acceptance. And that usually results in NEXTSTEP champions in the organization.
"Without the commitment and dedication of our technology staff and users, we wouldn't have succeeded as well as we did," says Brad Badeau, CFO of Toronto-based Trimark Investment Management, Canada's third-largest mutual-fund-management firm. "It's really important to have people on your side. Not every person will embrace change, but you must have a minimum for that embrace. You have to have more than one champion."
Trimark gained its champions when company executives watched a NeXT engineer build an image-management application in two weeks. They fell in love with black hardware and lobbied Oracle to port its Forms application to NEXTSTEP. Even through NeXT's transition, executives maintained their original commitment to hardware, snapping up used NeXTstations when they could be found, and set up their client-server network on Pyramid servers with 1GB of RAM and 50GB of storage. Trimark currently manages 550,000 fund accounts with assets over $6 billion Canadian.
"That commitment has worked its way through all levels of our organization," Badeau says.
A champion strategy has also worked well for the Val Verde Unified School District, in Perris, California, even though education typically has a "wider constituency" than business, says Darrell Lynn, IS director for the district. "When individuals find something that makes them work better and easier, they become champions. You can't afford not to have champions in all areas," Lynn says.
Val Verde administrators have just finished putting a NEXTSTEP machine on every teacher's desk in district schools that are wired with the FDDI (fiber distributed data interface) backbone to the district's T1 Ethernet system. Teachers and students are using NEXTSTEP in calculus labs and art classes, for yearbook production and Internet access, and as a districtwide curriculum server. An on-line help system and training classes help bring educators up to speed on the new platform. And the district held a month-long camp for sixth-, seventh-, and eighth-grade students this summer. "These kids are the superusers on campuses; they're going to be running classes for other kids and advocating. This is scalability of training," Lynn says.
No mystery train Training was cited by all users as the linchpin of their deployment strategies, though the level of training differs according to business needs. Jim McCrory, NeXT's manager of financial-services marketing, notes that "successful deployments have made an investment in training so users can take ownership of the new environment."
Alain Pinel found that even its veteran agents, some of whom have 30 years in the business and don't normally welcome change, have become "monsters on the system," according to Richards. With the help of Oakland, California-based Adamation, which designed much of the company's software, Alain Pinel has developed 27 training modules, each an hour in length. "There's always some training going on," says Richards.
The system has been so successful that Alain Pinel has networked four of its offices together to provide complete, on-line real-estate services to its clients, in-cluding forms management, feeds from listings services, customized listings complete with images of properties, and closing tracking.
The company has also brought local escrow and mortgage companies into the NEXTSTEP fold, providing a machine, software, and training to fully automate the purchasing process. "From the beginning, we saw the market as a big workgroup that spans company boundaries. Dozens of people are involved in closing a transaction: inspectors, escrow and bank officers, contractors. We're formulating a change in the processes of transacting, and that change is human rather than technological," Richards says.
For SBC, the very nature of its financial-trading business spans boundaries: The company has over two dozen offices around the world and maintains a sophisticated mainframe data-processing system, one of the world's largest private networks, and a huge legacy of technology investments made over many years, according to Koop. Thrown into the mix are agreements, contracts, legal descriptions, time frames, statistics, and analyses that are often site- and time-specific.
Given the frenetic nature of international finance, SBC opted for common functionality in its applications to specifically cut down on training time and foster cohesion among geographic groups. Because developers concentrated on a common look and feel across applications, traders were up and running in the new environment very quickly, Koop says.
While making sure that traders received the functionality they needed, SBC was also able to focus its training resources and prevent users' expectations from getting out of control by deploying its applications with a preconfigured workspace. This strategy has also worked well for McCaw Cellular, which has placed NEXTSTEP-based machines for activating customer cellular service in 50 service centers in test cities. The custom package was "designed completely for ease of use and much more customer satisfaction. Customers can get their phones activated in seven minutes instead of 30," says Ingvar Petursson, vice-president and chief information officer.
The system also provides on-line training, complete with a hypertext search engine and multimedia capabilities. The company plans to roll out 4000 machines for activation and customer service to 100 markets over the next year.
Chrysler Financial has also used a fixed-workspace approach in its custom rollout. To streamline training and get users quickly up to speed, it has also fine-tuned its deployment to a science. Every showroom that is set to receive Chrysler's new auto-financing system receives a training machine six weeks prior to the rollout date. Contractor Hewlett-Packard performs an infrastructure assessment at the showroom. Three weeks before the roll-out, HP engineers, working from a staging center in Corvallis, Oregon, get all the equipment together, including a custom-designed cabinet with outlets, fan, and uninterruptible power supply. They load the server, set up the master and clone for NetInfo, and configure network equipment. The machines arrive at the showroom on the Wednesday prior to the Friday deployment, when HP installs the PCs, printers, servers, new routers, hubs, and modems, laying new wire where necessary. On average, each site receives 25 machines; by the end of 1994, Chrysler plans to have 2500 machines in 100 branches.
For the first three days, trainers are on-site to answer questions about the custom apps, Mesa, WordPerfect, and custom calculators and screen savers. "Between the site survey, documentation, and training, we make sure the branch is emotionally ready for the rollout," explains Mike Adelson, project team leader.
Expect the unexpected In the end, even the best preparation, training, and involvement can't prevent the unexpected. Any major technological change in an organization will run into a few glitches.
For McCaw Cellular, the customer-activation system has been successful, but only as long as it's up and running. Petursson reports that because the company is "pushing the envelope" in client-server computing, the system sometimes crashes. So in designing its new customer-service centers, McCaw is spending additional time "simulating the user environ- ment" and testing all the components.
Another way to prevent disaster is to use every advantage you have. Williams Telecommunications, a large private-line fiber-optic service provider, perfected its NEXTSTEP work on its networks first, leveraging all its expertise before bringing the work to other areas of its business. Val Verde's educational mission has been carried over to training teachers and students to foster acceptance. Alain Pinel is a young company that could afford to completely redesign and construct a new computing system.
Alas, some disasters cannot be foreseen. When a bomb exploded at UBS's London offices in April 1993, the company lost windows – but its uninterruptible power supply prevented any loss of data. SBC's Chicago office, however, wasn't as lucky last April, when a water-main break shut down much of the city's corporate center. SBC sent its traders to London to continue working – and "learned to have completely ready hot backups," says Bill Martin, vice-president.
Eliot Bergson is managing editor of NeXTWORLD.