Local Area Networks Can Help Connect Busy Researchers

Last year, when biomedical researcher Henry Hsiao wanted to share data or test results with another scientist or engineer in his lab, he used to store the information on multiple floppy disks, then hand-deliver the disks to colleagues at the University of North Carolina, Chapel Hill. Similarly, when Hsiao, an associate professor of biomedical engineering, needed to have raw data processed by a large, number-crunching computer, he was required to physically take his data to one of the university's mainframes. But nowadays Hsiao can accomplish both of these tasks, and many others, with little more than a single keystroke on his personal computer. His PC is now tied into a local area network (LAN), a communications system for electronically transferring data among PCs, printers, plotters, and larger computers. "Before the LAN, we had a lot of transfers of floppy disks," Hsiao notes. "Now, using the LAN, I can run the computer in my lab from a PC I have in my office." LANs, once regarded as a technology affordable only by large commercial enterprises, have been available for more than a decade. But it has been only within the last few years that LANs have been making their way into university and smaller industrial research laboratories. In some of these labs, including Hsiao's, the networks have worked to reduce costs and increase productivity by allowing scientists to share software programs, data files, and expensive hardware resources, such as laser printers and powerful computational mainframes. In other settings, however, the networks have accomplished far less, primarily owing to poor planning and a lack of understanding about when a LAN should and shouldn't be installed. Put simply, a LAN is a communications tool. Generally, it consists of several microcomputers or workstations linked over a common cabling system, similar to a telephone network. Other LAN components include network operating software and a network driver, usually a dedicated personal computer on which this software runs. Depending on the type of LAN installed, a third component, known as a LAN server, may also be required. The server acts as a repository for all software programs and data files that users can share and exchange over the network. For example, if two researchers need access to the same software program, it is available to both of them simultaneously through the network server. This not only eliminates the cost of multiple copies of software programs, but also allows several scientists to work with the same data at the same time. In a networked setting, scientists can also share hardware, such as plotters or laser printers. In many cases, this is the chief reason a LAN is installed. But, unfortunately, it's not always a good reason, according to Wayne Rash, a networking specialist and technical director of American Management Systems, a consulting firm based in Arlington, Va. All too often, LANs are viewed primarily as a way of reducing hardware expenses, Rash points out. While the desire to save money on hardware is understandable, given the high price of lab equipment like plotters, using a LAN primarily for this purpose is generally not cost-effective, because of the even greater capital expenses associated with installing such a network. Rash says these expenses typically include the cost of network software, which can range from $2,000 to $20,000, depending on the network's configuration; approximately $5,000 for a LAN server; about $300 each for network interface cards, which must be installed in PCs and other equipment linked over the network; and, finally, the cost of wiring the room or building where the LAN is to be located. In some buildings, especially in those in which a new telephone system has been installed, it may be possible to connect the LAN over extra twisted-pair wires located in many of the newer phone wiring plans. In older buildings, a new LAN-specific cabling system - based on either twisted-pair wires or more costly coaxial cable - is usually necessary. To get the most out of the technology, scientists must bear in mind that LANs are designed to enhance communications, and not to replace hardware. Therefore, one of the first tasks in deciding whether a network is necessary is clearly defining the communications requirements of a laboratory or organization. Do researchers need to transfer data quickly to other workers? If so, would transferring data electronically improve productivity? Do workers need access to information at the same time? Will the network be used primarily to share data files, rather than for sending memos and other electronic mail messages? Does information need to be sent to other workers whose personal computers are already networked? If you answer "yes" to two or more of these questions, chances are your lab could benefit from a LAN. Answering still other questions will help you decide what type of LAN is best for your lab. How many workers will need network access to printers and plotters? Will they need access on a full-time or part-time basis? How far apart will workers on the network be located? On the same floor? In the same building? What kind of traffic will the network carry? Primarily graphics or alphanumeric data? There are several types of LANs from which to choose, ranging from lower-cost, packet-based systems, which can operate over conventional telephone wiring, to heavier-duty multiple-channel, broadband LANs, which are more suited to heavy graphics traffic. Choosing the right network will depend largely on your answers to these questions. If an organization is to get full benefit from a network, all of these questions must be addressed. If they aren't, a LAN may turn out to be no more than an expensive cat's cradle of wires, which is of little practical use. Notes Rash: "The biggest danger people run with this technology is installing a LAN and then trying to find ways to justify its installation." At Hipple Cancer Research Laboratory, scientists were spared the task of cost-justifying the LAN they have had up and running for more than a year, because the network was donated to the facility. The LAN was given to the research center by 10NET Communications Corp., a manufacturer of LAN systems, which, like Hipple, is located in Dayton, Ohio. Currently, Hipple has all 18 of its personal computers networked over the 10NET system, which Susan Petrose, Hipple's computer systems manager, says has increased productivity several times over. Essentially, the networking system allows scientists to access the programs and memory of all PCs throughout the laboratories. As a result, scientists can compare research data and send reports to the center's document processing department, without ever leaving their labs. Petrose says that in addition to saving time, the LAN has also saved the center money by allowing scientists and administrative personnel to share hardware and programs. For example, Hipple's scientists, postdoctoral fellows, and technicians can produce graphs at their own computers using programs like Sigma Plot and Harvard Graphics. They then can ship their graphs over the network for print-out at a plotter, also tied into the LAN. Petrose says the LAN has proved particularly useful to the staff of the International Journal of Cell Cloning, which is published by the center. The LAN allows the center's scientists to review articles and submit their comments to the journal's editorial staff, which in turn can share editing duties on the same manuscript because it can be called up simultaneously by any of the networked PCs. Also, print-outs of manuscripts can be sent over the network to a single laser printer, located in a central processing department. For those organizations that may want to experiment with a LAN on a small scale - perhaps in a single laboratory - before undertaking an enterprise-wide networking project, there are relatively low-cost LANs available for such departmental applications. In recent years, several LAN companies have developed products for small to medium-sized installations as well as for individual workgroups within large organizations. In addition to 10NET, these companies include the Tucson-based Artisoft Inc., and Invisible Software Inc., of Foster City, Calif. Both companies offer LAN systems designed for networks of fewer than a dozen PCs. Neither system requires a dedicated LAN server, making such networks less costly and easier to install than other LAN systems designed for larger groups. Julia King is a freelance science writer based in Ridley Park, Pa.

Local Area Networks Can Help Connect Busy Researchers

Interested in reading more?

Become a Member of