Casting your net for optimal networking.

Networking has become a buzzword that buzzes beyond its actual meaning. Magazine articles, and not only in the computer field, devote many pages to networking systems of all kinds. Computer companies rush to make their systems "network friendly" or compatible. People network with each other for career advancement. This article will discuss networking in the context of the clinical laboratory.

The general principle behind networks is to connect computers in some fashion to share parts or resources. A network consists of one central computer, usually called the network server, and several peripheral computers and/or terminals. The central computer may play both roles.

At a minimum, each computer consists of a keyboard and a monitor, sometimes called a "dumb" terminal because it has little power on its own. Computers are connected to hard disks, printers, and other hardware. Computer software may run in one computer or all.

Networking may be essential or merely useful. You must balance your needs and benefits versus the costs in terms of money and headaches.

Networks have two main applications: to share data (results or data files) and to share resources (hardware or software).

* Sharing data. The main advantage of a network is that it permits many users to have prompt access to up-to-date data, which may consist of laboratory results or data files such as pathology reports. "Concurrent" access is really a misnomer, since the sharing of information does not happen instantaneously. Data are shifted so quickly among users that each user appears to be using the system alone.

When many users attempt to use the system at the same time or make heavy demands on it, however, it slows down. There are other disadvantages as well. Having a network requires installing a system of increased complexity and usually of higher cost than is already in place. The introduction of a network brings with it the uncomfortable possibility that the entire system will collapse when the central computer is not working.

In the past, vendors would sell a cheap computer that worked well during demonstrations. When it turned out to be inefficient for the laboratory's requirements, the same vendor would sell the same user a more powerful machine due to "unanticipated" increases in computer utilization. While this problem is less prevalent today, any network can become overworked if too many people use it.

The two greatest benefits of using decentralized computers are their relative simplicity of operation and their ability to continue to operate even if one computer fails. The two main drawbacks are that real-time (up-to-the-minute) file updates are not possible and that data sharing among many users may be complicated.

Modern software and hardware permit concurrent use by many users, each with a different level of authority or accessibility regarding which data can be viewed or altered. Cost and speed depend on the extent of security and checks imposed on the system. If a central laboratory produces copies of its database for distribution to local computers, users of any of the local computers cannot modify files from the central computer. At worst, they can damage their own copies, which are easily replaceable. They cannot, however, access the latest data.

* Sharing files. A network is useful for sharing files containing documents that require frequent updates or modification. Residents, pathologists, and secretaries type numerous reports on anatomic and clinical pathology, for example, and various form letters to physicians and patients. The staff dictates reports, which the secretaries transcribe.

In many labs the secretary types a new report for each patient even though similar reports were typed for many others. It would save considerable time and energy for the secretaries and staff to be able to access a set of files consisting of standard reports for commonly found conditions. One solution is a network in which all terminals can access a central hard disk containing all such files. If the same document is to be edited by different people, it may be necessary to have a central hard disk to assure that each person works with the latest version.

Such a system is easy to implement. Separate terminals are installed in the resident's room and in the offices of key pathologists. A central computer contains all standard reports as well as reports on all patients.

Here is an example. The resident dictates a report and refers the secretary to a specific report format. The secretary transcribes the dictation by filling in the information. The resident proofreads a draft on a computer screen and makes corrections. The pathologist reviews it on another screen and makes additional corrections. Under most current practices, the resident tapes the report, which is transcribed by the secretary, who prints a draft for review by the resident and pathologist. The secretary inserts the corrections and prints a final report.

It all works well when everyone is present at all times. If the secretary is not available, however, the report waits. With a central file, any secretary could access and correct the report. Problem: Errors could multiply, as anybody might mistakenly "correct" other people's reports. To prevent this, a password must be assigned to each report, which may be edited only by designated persons.

Having a network to share files is important in an office where different users write similar letters or documents and want to access a set of similar files, not necessarily at the same time. Backup procedures are simplified because the caretaker of the network routinely backs up the central hard disk. By controlling access levels, network programs can prevent snooping.

* Sharing software. Commonly used computer programs may cost $200 to $500 per copy. Some highly specialized programs cost thousands of dollars. Vendors require that a separate copy be purchased for each computer on which it is used. With a network, all computers can access one copy of the software--in which case it may be necessary to have a version designed especially to be run with networks. Although such versions are usually more expensive than a single user copy, they rarely cost more than twice as much. If three or more computers use the same software, money can be saved.

Saving money on software used by all computers is a legitimate reason to have a network. Before you choose the network option, consider the possibility of buying site licenses on software; these allow multiple computers to utilize the same software. Perhaps computers could be designated for certain uses and not others. Some software companies allow you to copy the software on several machines as long as it is not used on more than one at the same time. For example, if you have two computers, one at home and one in your office, you may be able to use the same statistical program on both at different times.

Software companies differ widely in their requirements regarding the copying of programs. Some companies not only prohibit this multiple use but require installation of a special gadget on the printer port or elsewhere before the software will work. This gadget is an electronic "key" that allows the software to run. To use the statistical program at home and in the office, the user must carry the key back and forth. Although the gadget is small and easy to install and remove, the software will not work without it.

Beware: Some software does not work well with networks at all. Always check with your software vendor before attempting to run the software in a network, with multiple printers, or through printer-sharing hardware or software. This is especially important regarding software with copy protection.

* Sharing hardware. Sharing expensive hardware resources used to be, and perhaps still is, among the most popular reasons for installing small networks. A group of computers may share a laser printer, a scanner, and other expensive hardware such as a color plotter, backup tape, and more.

Networks to share hardware are easy to set up and require minimal maintenance. Because hardware is shared in only one direction--from PC to hardware--there is little opportunity for any user to modify somebody else's files. Devices exist that will shift the printer from one computer to another, either manually or automatically. A future article will discuss various printer buffer options.

* Do you need a network? Having a network in place is only one of several ways to share expensive hardware or software. Other methods include site licensing of multiple copies and dedicating computers to specific tasks. Expensive cable installations may be required to network computers in distant locations.

If you do not need to share a central database of numbers or text files, it would probably be better to have separate but similar computers dedicated to different tasks. In an emergency, any computer can substitute for another.

* Alternatives. It's handy to have several independent but similar computers that can perform specialized functions but are interchangeable in a pinch. Given the substantial reduction in computer prices evident in recent years, you can purchase a fast AT clone with an 80286 chip[1] for approximately $1,000. This computer would be sufficient for most word processing needs.[2] Several of these computers can share a laser printer.

Site licenses of commonly used word processing software are available for about $100 per copy. A faster computer can be bought for graphics, data analysis, and database applications. This computer also shares the laser printer and may be used for word processing, if necessary. There is no need for a network.

It is possible to start with a group of computers and network them later. If this approach is used, consider buying computers without hard disks that are very large (more than 100 Mb) and very expensive. The central computer may later provide the storage space you need.

Our laboratory has several computers in widely separated locations and used by different people. I installed a card (board) in each computer and can move my tape from one computer to another to make backups.

I purchased several inexpensive laser printers to be shared by the computers that are located near each other. To print a file, users copy it in a floppy disk and carry the disk to another computer attached to a printer. The extra time required is small in the context of file preparation.

* Not an office. Sharing data, hardware, and software is almost essential in the laboratory. We need a central database, yet software is often so expensive that our budgets forbid the purchase of multiple copies. Networking is frequently the answer.

Nevertheless, laboratories by their nature are very different from business settings. It is not feasible to shut down all terminals in a laboratory for extended periods while someone plays with a new networking program. Laboratories deal with critical data and must maintain a high degree of accuracy at all times.

Many networking options appropriate for a business office would not be usable in a clinical lab. While it's fun to get excited about articles in computer magazines, make sure the network system chosen for your laboratory will suit your individual needs.

[1]Siguel, E.N. Which PC to buy now (Computer Dialog). MLO 22(9): 59-61, September 1990. [2]Siguel, E.N. Software choices for today (Computer Dialog). MLO 22(10): 59-62, October 1990.

Edward N. Siguel, M.D., Ph.D., J.D. is a senior scientist in the clinical nutrition unit at University Hospital, Boston University Medical Center, Boston, Mass.