Making Science Make Sense
| Home > Scientific Writing and Editing > Feature Article Sample | Home | About | Samples | Contact |
|
Harnessing the Hype: Building Successful Discovery Applications on the Web
Yet the presentations and conversations at the User Conference also peeled away the layers of hype obscuring some of real challenges involved in Web development. The Web, in fact, is not always the right tool for the task. And when it is the right tool, careful planning and an understanding of the quirks of the Web environment are essential to success. Since Chemscape's release three years ago, MDL's developers and consultants have learned firsthand the pros and cons of working with the Web. They understand why the environment is so appealing, but also know from experience some of the common pitfalls that can trap novice Web developers. Molecular Connection talked to some of these individuals to get their take on what works -- and what doesn't -- when companies start building discovery applications on the Web. Beyond "Cool": Why Business Likes the Web The Web has captured the public's imagination more than any other technology in recent years. Perhaps this is because, unlike other development environments that seem foreign and inaccessible to the average user, the Web has a personality. People who would never consider picking up a book on C++ readily try their hand at HTML and JavaScript. The Web is graphical, creative, expressive: in a word, "cool." Just ask any serious Web developer; inevitably, the conversation will include the word "cool." "The Web is hot right now, and with good reason," said Jean Holt, marketing scientist for core technologies at MDL and a key force in MDL's Web strategy. "It can do cool stuff -- rotating graphics, hyperlinks, the ability to jump to information stored in very different locations. And it satisfies IT people, who want to do something one place and distribute it everywhere rather than updating DLLs at thousands of computers." The Web certainly offers a deal from a business perspective. Browsers are free; hosting and Web access are cheap; and software, upgrades, and documentation are maintained and distributed centrally. From a common user interface, scientists can access text, images, video, and sound from any type of computer platform. "Think about it," notes Holt. "To distribute chemical information worldwide, to some extent all you need on scientists' desktops is a browser and the Chime Pro plug-in." In addition, people know and like the Web. Scientists browse and surf the Web in their spare time, so bringing browser technology to on-the-job applications means one less interface for scientists to learn. According to Holt, IT groups have even found that prototype testing works easier on the Web, as users are more willing to experiment with Web applications than other types of software. The Web's inherent "coolness," combined with its business value, has propelled organizations to move all sorts of information management to the Web. At the 1999 User Conference, customers demonstrated chemical and biological registration applications, inventory and compound management systems, and full-blown discovery systems for accessing all types of scientific data throughout the enterprise. While these large-scale applications tend to steal the limelight, some of the best applications involve focused use of Web functionality. Mitch Miller, a senior consultant at MDL who was part of the MDL team that first released Chemscape in 1996, has helped complete several major projects using the Web to access discovery information. It's a role that has let him see some of the unqiue utilities that only the Web makes possible. "Intelligent hyperlinks can really help to build integration with other systems," he noted. "Graphical links and widgets also work really well." "All discovery applications in the life science and chemical industries are suited for the Web," explained Dr. Dan Fercu, a senior applications consultant in MDL's Midwest U.S. consulting office. "Web functionality can help researchers manage the maze of cross-references and information from different sources inherent to discovery, exchange ideas, and cooperate in their research." But First, Are You Experienced? While all applications may be suitable for Web development, some applications are more suitable than others. "The Web undoubtedly has the power to extend access to scientific information within an organization and to enable researchers to view in-house information in a broader, worldwide context," Miller said. "This can tempt organizations to put a Web face on every application, even though, in some cases, the tools are still evolving." Holt agreed. "The Web is universally loved for its versatility and its ability to quickly and easily link to many related data sources," he said. But this very interactivity and flexibility means that Web-based applications can quickly mushroom, with developers taking on more than the environment can handle. "Because we are dealing with our customers' intellectual property and perhaps an organization's next billion dollar molecule, we can't be casual about it," Holt continued. "It's up to developers to rein in the Web's inherent, free-wheeling environment, while still providing the type of easy access and maintenance that has made the Web so popular." Trying to do too much is one of the most common mistakes that companies make the first time they develop on the Web. In particular, Miller suggested caution for organizations that want to replace functioning client/server systems with Web-based ones. "Be sure to look closely at functionality," he advised. "Client/server systems have easy access to user-specific directories on both workstation and host computers. While you can definitely provide that type of access on the Web, it often requires additional planning and development. Good judgement should prevail, with groups working to understand researchers' needs fully to determine what can be delivered expeditiously with Web technology and what is better served today using other tools." The popularized simplicity of the Web environment also causes problems for developers. Fercu noted, for instance, that inexperienced developers often fail to consider the Web's three-tiered architecture: the browser for data presentation, the application server for business logic, and the database server for storage. Capitalizing on this architecture means understanding its limitations as well as its benefits, according to Holt. "On the Web, performance is entirely dependent upon your Web server and your host communication," Holt said. "If the server load is too high, performance suffers." In addition, downloading too much data at one time can often crash browsers. That's why the best Web applications today have a well-defined scope and purpose; controlling the way scientists interact with the application not only makes their jobs easier, but improves overall system performance. "Most companies are attracted to the Web because of the control it offers in terms of maintenance and updates, but when it comes to developing the application, programmers simply do not have the same amount of control as they do with a standard Integrated Development Environment," Holt explained. "You do not have as much control over the actions performed by your users, you have little control over the errors that are generated, and, worst of all, there is a major browser battle. Something developed for Internet Explorer may not look the same or even work in Netscape or Opera or HotJava." To combat technology problems, Fercu recommends that companies stick with one version of the software until the application is finished. "Trying to support different Web browsers or different versions of the same browser, or upgrading background applications midstream can stall development," he said. Because the nuances of Web technology make it quite different from traditional development, Holt recommends that companies hire people with direct experience developing on the Web. And this doesn't just mean software engineers -- graphic artists are also essential. "Visual appeal is key to usability on the Web," Holt explained. "Software engineers may be able to draw, but they don't always know how colors work together and how to make a page come together. Artists are necessary to make pages that scientists will be happy to use day in and day out." The Power of Planning The Web has certainly grown into a viable environment for application development. But before jumping on the bandwagon, MDL's developers and consultants recommend that organizations plan, plan, and plan again. Fercu explained that for a recent consulting project, thorough planning kept the project on track and well in line with user expectations. "I created a mock-up of the application on my blackboard to get a feel for the dynamics of the links within the application," he said. "Then I used template pages with no real data but with all redirection information included to check how the links really worked. Only after these two preliminary conceptualization steps did I move ahead to the production and testing stages." Holt concurred. "Successful Web development starts with a lot of questions to help you assess the objectives and scope of your application," he said. "Is your application serving 10, 200, or 1,300 people? What is your network like? How complex are the tasks your users are engaged in? How quickly can you modify Web browsers? You may find that the Web isn't the best place for your application, but you will only know for sure by examining carefully where you are today and where you want to go." Every experienced application developer knows the value of looking before leaping -- of following good development practices to bring concepts to reality. Web development is no different, if you listen to MDL's successful developers, consultants, and customers. Assess the terrain, measure the distance, then go ahead and leap. |