Special Report: Software That Links Distant Researchers

For the many researchers who are comfortable with and knowledgeable about computers, shopping for communications software can be a pleasurable and educational experience. Yet for those who know little about computers, the task could be painful, if not expensive. These scientists would do well to first understand the whole process of how computers communicate. Scientific computing used to be largely confined to mainframe devices, with all necessary data, operations, and report generation per- formed by a single machine. But with the explosion of personal computer technology that has taken place during the last 10 years, communication between computers has become important, if not a necessity, to many scientists' work. Researchers now can discuss data and concepts via national and international computer networks. A scientist with a personal computer can send the latest draft of a research paper to a colleague in the next room--or the next state. Reams of data are available from faraway labs through online databases. Of course, in order for all these electronic interactions to take place, software is needed to tell the computers how to talk to each other. Before any communication between computers can occur, the machines must be physically connected in some way. If the computers are at the same site, they can often be connected by cabling the machines together--a procedure called "hard-wiring." If the computers are too far apart to be hardwired, but are still in the same building or lab, they can be linked--more or less permanently--by way of a local area network (LAN). A LAN usually consists of several personal computers connected to each other, and often includes a more powerful computer that holds shared data and programs (The Scientist, Feb. 5, 1990, page 31). A network operating system is used to manage the linkages and the communications traffic among the computers. Users connected by a LAN can communicate quickly and easily using electronic mail software, or E-mail. They can send brief messages to each other through their computers, without walking down the hall or interrupting a colleague with a telephone call. While LANs are valuable systems for a confined area, such as an office, lab, or building, most intercomputer communications involve longer distances, and are handled with temporary linkages over the phone lines. To accomplish this, a separate piece of hardware called a modem is necessary. The term comes from "modulator-demodulator," which describes what the modem actually does; that is, to modulate, or change, the electrical output of a computer into an audible signal that can be sent along the phone lines, and then demodulated back into an electrical signal for input to the receiving computer. Most communications software packages include the capacity to specify the characteristics of the modem in use; the most important characteristic is the speed of transmission. Speed of transmission is of considerable importance when a user is transferring large amounts of data, particularly over a distance, as the per-minute cost of maintaining a phone link may be high. The maximum speed of a transmission is limited by the capacity of the modems involved. All communications packages allow the user to set modem and phone-link parameters. Most will also provide amenities like built-in phone directories for most-called numbers; auto-answer (the computer answers the phone when it rings, and receives any incoming data transmissions automatically); self-diagnostics; options in capture method (whether incoming data is stored in RAM memory, or directly to a disk file, for example); and time-set automatic file transfers, so that one need not be physically present in order to have the transmission take place. The most frequently used type of communication between computers is file transfers. A "file" may be numerical data, text, graphic images, or a mixture of any of these. The contents of a file make no difference to the software that transfers it from one machine to another--to computers and to software alike, a file is no more than a stream of electronic bits. "Protocol" is a term one frequently encounters when shopping for communications software. A protocol is the particular way in which the software handles data transmission, and any one communications package generally includes a choice of several protocols, including simple ASCII file transfer, which is essentially "sans protocol." The cost of a package tends to increase along with the number--and sophistication--of protocols it includes. There are several popular and commonly used protocols, including Xmodem, Zmodem, and Compu-Serve B. Xmodem is a fairly old protocol, introduced more than 10 years ago. It functions by sending a block of data, then waiting for acknowledgment that the block has been received before sending another. Over the last few years, "streaming" protocols have been developed; these send data continuously, simultaneously listening for acknowledgment of earlier data from the receiving computer. The Zmodem and CompuServe B protocols are streaming protocols. In order to take advantage of the attributes of one of the more advanced protocols, both sender and receiver must be using it when a file is transferred. File transmission can be speeded up through data compression, an option offered by many communications protocols. Data compression means that the file to be transferred is divided up into "blocks" of bytes, in a manner that eliminates as much blank space in the file as possible. Hence, the data can be transferred in less time than uncompressed files. While this is generally accomplished by software control, some modems are capable of doing data compression without compression software. For this to be effective, however, the modems on both ends of the transmission must have this capability. When data are sent over phone lines or satellite links, there is always some danger of contamination of the data by "line noise"--that is, random electronic impulses generated by the equipment or picked up from the environment. Particularly bad lines can cause enough noise to corrupt a data file past the point of usability. In addition, data compression sometimes introduces errors into a file. The more sensitive the data being transferred, the more important error correction becomes. Communications software corrects errors by dividing the data to be transmitted into blocks of a certain size and then attaching a mathematical "check-sum" to the end of the block. When transmitted data are received, the receiving computer's software will check the result of the check-sum for each block as it is received, and transmit the result back to the sending computer. If the returning check-sum result is the same as the check-sum that was sent, it is assumed that the data have arrived without error. If not, an error is assumed to have occurred. Communications packages that compress data also include error-checking capabilities. Depending on the software involved, the block of data with the error(s) may be re-sent, or the software may merely inform the user that an error occurred, and leave it up to the user to decide whether to resend the file or data block. Error correction is obviously desirable, but has a price, in that it takes longer, thus causing slower (sometimes much slower) throughput of files. When dealing with enormous data files, error correction can involve very large--and expensive--delays. Hence, modem speed and data-compression capabilities become quite important when large files are checked for errors as they are transmitted. Software packages handle error correction in a variety of ways. Older protocols such as Xmodem send an entire file, wait for acknowledgment, and then re-send the whole file if it contains an error. With slightly more sophisticated protocols, after the file is sent, just the blocks that contain errors are re-sent after acknowledgment. Streaming protocols can re-send blocks with errors without waiting for the entire file to be sent. In addition, some software packages allow the user to select either full error-checking and error-correction or an optional, faster mode, in which no acknowledgment is required from the receiving computer, and only one check-sum is done, at the end of the file, rather than at the end of each data block. A more advanced version of streaming protocols is the recently developed "smart" protocols. These are ideal for use in updating large databases, in that a "smart" protocol will send only those portions of a file that have been changed since the last transmission. DCA/Crosstalk Communications of Alphoretta, Ga., includes its DART protocol with the company's Crosstalk Mark 4 package. DART is a streaming protocol that checks the receiving computer for an extant file and sends only those parts of the file that are different in the sending computer. Several protocols, including Zmodem and CompuServe B, can resume an interrupted file transfer without the necessity of starting over from the beginning of the file, but these don't do the "new-parts-only" transfers. Besides error-checking and data compression, some communications software has the ability to encrypt data. This is not necessary for most data transmissions, but may be desirable in the case of very sensitive data. Choosing a protocol depends somewhat on the amount and destination of the information being sent. A simple protocol, such as Xmodem, is sufficient for sending short messages or files. But for complex tasks, such as the constant updating of an enormous database, more sophisticated protocols are needed to expedite the process. Online networks such as BITNET are compatible with common protocols such as Xmodem and Zmodem, as well as with straight ASCII file transfers, but they don't involve data compression or error correction. Finally, "no protocol" is sometimes the best protocol. An ASCII transfer doesn't include error-correction, but is compatible with virtually all systems and software. While file transfers are the major functions of communications software likely to be useful in scientific research, some communications packages offer other helpful features. Close-Up software, available from Santa Barbara, Calif.-based Norton-Lambert Inc., for example, allows the user not only to do file transfers, but also to operate another computer as a "remote" site, controlled from one's own keyboard. This is particularly helpful for those with computers both at home and at the office who wish to do automated procedures from home; they can leave the office computer turned on when leaving, and then perform their procedures from home, treating the home keyboard as though it were the keyboard of the remote computer. With laptop computers so popular these days, the problem of file transfer takes on another dimension. Often, a researcher may need to do work on a laptop while traveling, then transfer the results to a larger machine at the lab or office. While file transfers can be done between a laptop and another computer using modems and communication software, there is a less expensive alternative. LapLink III, by Traveling Software Inc. of Bothell, Wash., provides a cable for connecting your laptop to a larger computer, and a software program that allows the laptop to be treated as an additional drive to the larger computer (or vice versa). Thus, files can be transferred without the necessity of a modem on either machine. A computer equipped with communications software and a modem can access any of several large databases that provide information and communication with other scientists. There are literally hundreds of commercial databases available, supplying information on such topics as chemical reactions, pharmaceutical registrations, geographical data, astronomical data, and so forth. Most of these are accessible through a central commercial database service, which can be dialed into with a modem. There are several such services; DIALOG Information Services, based in Palo Alto, Calif., is one of the largest U.S. database suppliers. In addition to online database resources, computer users belonging to a university or other large institution can usually obtain access to BITNET, a very large communications network linking a great number of academic users. Over a BITNET linkage, a scientist can converse--at a relatively low cost--with colleagues all over the United States. Great Britain has a similar system, called JANET, which maintains linkages with BITNET. Messages and data can be transmitted easily over both systems. Another widely used online service is CompuServe, from CompuServe Inc. of Columbus, Ohio. Unlike BITNET, which is academia- and research-oriented, CompuServe has a fairly broad scope, providing businesses and individuals with such services as bulletin boards, online conferencing, electronic mail, and travel and entertainment databases. Very little sophistication is required from communications software for access to online resources. Virtually any package can dial into an online service, and once the modem link is formed, most online operations are carried out by the software belonging to the online service. The computer revolution is part of the wider communications revolution. With the computer communication options now available, databases, bulletin boards, supercomputers, and colleagues are just a modem and a phone call away, bringing scientists a big step closer to a global community of peers. Diana J. Gabaldon is assistant research professor at the Center for Environmental Studies, Arizona State University, and editor of Science Software.

Special Report: Software That Links Distant Researchers

Interested in reading more?

Become a Member of