Library Technology Guides

---

Volume 2 Number 07 (July 1982)

Assessing telecommunications options for minicomputers

Most of the discussion on the computer usage implications of the restructuring of the telecommunications industry have addressed the issues in the context of the mainframe environment. However, since the majority of automated library systems are mounted on minicomputers, the editors offer the following discussion of telecommunications options from that perspective.

The data communications needs of minicomputer-based systems are significantly different from those of mainframes. For example, mini-based systems are typically concerned with communication over relatively short distances, typically under 100 miles, in contrast with the hundreds or even thousands of miles common to mainframe communications. Additionally, mini users usually require a limited number of linkages rather than a complex multinode telecommunications network. Finally, of particular significance to telecommunications, the terminals on mini-based systems are usually asynchronous "dumb" terminals, whereas a mainframe tends to be linked with synchronous terminals with some built-in communications protocols.

Communications protocols, or standards, such as IBM's BISYNC, permit terminals to be "multi-dropped" or grouped on a single telephone line because each terminal can be selectively "polled" or addressed by the computer. Minis are typically configured with only one terminal per computer port so that the computer does not need to ascertain which terminal is transmitting, and therefore normally do not have this feature.

A "dumb" terminal uses no communications protocol. It is a Teletype® or Teletype® compatible terminal that displays or prints data just as it receives it. A "dumb" terminal may incorporate a microcomputer for data reformatting and other local activities, but it is still "dumb" if it is Teletype® compatible. Without a protocol a terminal cannot be polled, so it cannot be clustered with other terminals on the same telephone line unless linked to additional equipment which contains a communications protocol.

The mainframe protocols exist because manufacturers have had enough users with numerous terminals distant from their mainframes to warrant the development of protocols by the vendors. Most minicomputer-based systems use "dumb" terminals which are less expensive and completely adequate for the tasks to be accomplished, because so many of these terminals are located within a few hundred feet of the computer site and can thus be hard-wired to the computer. Unfortunately, when there is a need to link remote terminals, the options are limited because there is no standard communications protocol to adopt. Terminals in the same building as a computer are usually hard-wired to it over a simple twisted pair of wires using line drivers-signal converters which condition the signal to ensure reliable transmissionon each terminal and the associated computer ports. Twisted pair wires may be installed by the telephone company, the computer system vendor, or the system user.

Remotely located terminals, those that cannot be hard-wired, are usually connected to the minicomputer by a telephone line with modems at each end. (The Bell System uses the term data set rather than modem.) A modem or data set is an electronic device which converts the signal from the digital form used by computers to the analog form used for telephone voice transmission and vice versa. The signal is modified to appear to the telephone system as a voice communication and to the mini computer as a digital communication. A pair of modems usually costs $1,000 or less.

There are three common methods of establishing the telephone linkage: dial access, dedication of a leased line to a single terminal, and the sharing of a leased line among a number of terminals using multiplexing techniques.

The dial access or dial-up approach enables regular telephone installations to be used for data transmission. The line can be used for voice communication when not being used for data transmission. Dial access is appropriate for situations where a specific terminal-computer linkage is of relatively short and infrequent duration-entailing only a few dozen transactions a day. In an environment where a number of terminals need to access the computer on such a basis, it is possible to have more than one terminal assigned to a computer port or input channel, thus reducing the cost of the computer installation.

The efficiency of dial access is reduced by the limitations which apply to regular telephone traffic-peak period loads which can result in difficulty in establishing the connection-and restricted transmission speeds. Until recently, a major advantage of this method of terminal-computer linkage was economic. Most such installations are within local areas where relatively short distances are involved and local calling rates apply. This situation is changing as telephone companies redesign their pricing structure and charge business and institutional users by The minute for local telephone calls. Costs are expected to rise further as rates are restructured to reflect the loss of long distance traffic which currently and in the past has helped to subsidize local service.

Certain technical aspects of dial access linkage limit the extent to which techniques to increase the speed of data communications-and thus lower the costs-can be applied.

Transmission speed is usually expressed in "baud" one baud being one signal element per second or bits per second (bps). With voice-grade telephone lines and regular modems the practical upper limit is 1800 baud or 1800 bps. Higher speed modems are available. They achieve faster data transmission by encoding more data bits in a baud-a modem operating at 1200 baud, but encoding two bits in a baud, effectively transmits data at 2400 bits per second (bps). Some modems will transmit data over voice-grade lines at 4800 and 9600 bps. Unfortunately, another characteristic of "dumb" terminals is that they can only use full- duplex modems in dial-up mode, and this effectively limits transmission to the basic 1200 bps.

Given the changes in dial access pricing structures and an appropriate number of transactions per remote terminal per day-300 or more- most libraries are in a situation that justifies the use of a dedicated leased line. The cost of such service now varies from $4 to $12 per month per mile, but it can be expected to rise dramatically in The next few years.

Although in many parts of the U.S. the phone companies offer Dataphone Digital Service (DDS) as an alternative to voice-grade lines, most leased lines are voice-grade. With present tariffs, there are virtually no cost savings to be realized through the use of DDS even though no modems are required on equipment attached to such lines. There is, however, greater reliability and better performance.

Terminals located in groups at remote sites may share a leased line by being "multiplexed" to a minicomputer using data terminal equipment designed to combine the transmissions of multiple terminals into one composite signal. At the other end of the line on which this signal is transmitted, an identical multiplexor reconstitutes the original input from each terminal and routes them to the appropriate ports of the computer or to yet another multiplexor. Neither terminal equipment nor the computer hardware or software need be changed when multiplexing is undertaken. Dial-up lines may also be shared this way.

There are several multiplexing techniques, the most cost effective of which is usually statistical multiplexing. A statistical multiplexor uses a small microprocessor and a buffer memory so that data can be stored temporarily during periods of peak activity. This permits more terminals to share a line because the "stat mux," as it is often called, smoothes out the traffic flow. It allocates the shared line in such a way that up to eight terminals, each operating at 1200 bps, can share a single 1200 baud voice-grade telephone line transmitting at 2400 bps. Another term now more commonly used by manufacturers of this equipment is "data concentrator." Stat muxes or data concentrators have a price range of $1,500 to $10,000 each depending upon the number of terminals they handle. In addition, the special modems required for the data concentrators may cost as much as $6,000 a pair.

To add to the complexity of all this, it is possible to network stat minces or data concentrators. For example, several terminals at a branch library may share a multiplexor (unit A) that is connected to another multiplexor (unit B) through a pair of modems. In turn the second multiplexor (unit B) connects with yet another multiplexor (unit C). Several other terminals and terminal clusters may also come into unit C so that all share a single line to the central site. The multiplexor (unit D) at the central site splits all the transmissions among the appropriate ports of the computer.

In addition to the equipment already mentioned, there are concentrating terminals, multidrop concentrators, port concentrators, and intelligent port selectors which further stretch the number of telecommunications options. These will be pursued in a future issue of LSN. Given the range of the options available to minicomputer users and the expected increase in the number of suppliers of telecommunications hardware and transmission lines as the telephone industry is deregulated, librarians managing automated systems will need to periodically reexamine their telecommunications approaches to ensure that they continue to take advantage of the most cost effective options available.

The editors offer the following rules of thumb as valid for the time being:

1. If the terminals are within 2000 feet of the computer, hard-wired or direct connections using line drivers are generally most cost effective.
2. When terminals are remote from the computer, but widely scattered, direct connection through telephone lines with modems may be cost effective. Normally dedicated lines are better than dial-up. If a library' s telecommunications costs are more than $1,000 a month or have risen more than 20% in the past year, a review should be undertaken.
3. If the remote terminals are concentrated at a small number of sites; there are more than 25 remote terminals; or telecommunications costs are more than $1,500 per month, it is quite likely that the use of statistical multiplexing will be cost effective.
4. If there are more than 50 remote terminals, it may be possible to network statistical multiplexors to realize even greater cost savings.

Some libraries may be able to reduce ongoing telecommunications costs by up to 90% by investing in telecommunications hardware. The "payback" period--the time required, to recover the capital outlay for telecommunications hardware-may be as short as two to three years.

While a library may wish to have the vendor of its computer system review and modify its telecommunications, it is not mandatory that it do so. Virtually all telecommunications hardware requires no changes in software. A library may choose to retain an expert in telecommunications because its vendor does not have appropriate expertise or because the vendor's prices are felt to be too high.

The vendor should be notified of planned changes as should the telephone company (ies). Despite frequent protestations by telephone companies, they may not deny the connection of telecommunications equipment to their lines. Until 1968 only telephone companies could connect telecommunications or other devices to their own facilities. As a result of an FCC ruling known as the Carterfone Decision, independent data and voice communication vendors may now attach a wide variety of devices on behalf of their customers.

Online Computer Systems actively marketing ILS

Online Computer Systems Inc.'s recent ad in Special Libraries served notice that it is still actively marketing the Integrated Library System (ILS) developed by Lister Hill Laboratories for Biomedical Communications. The company says That it has more ILS sites and more ILS experience than any other company. It supports the ILS installation at the Army Library, Pentagon, and has recently sold systems to the University of Maryland Health Sciences Library and the U.S. Naval Research Laboratory Library.

While the software is owned by the United States Government and is made available to the public through the National Technical Information Service (NTIS) for approximately $2,000, NTIS does not provide hardware, operating system, installation, training, modification or ongoing support. Online Computer Systems and the other ILS vendor, Avatar-provide these additional services. The minimum cost for a complete turnkey system for a small library is $50,000. [Contact: Online Computer Systems Inc., 20010 Century Blvd., #101, Germantown, MD 20767. (301) 428-3700.]

Appleton Public Library selects CLSI

The Appleton, Wisconsin, Public Library has become the first library to order CLSI's new microprocessor- based System 23. The System 23 employs a Digital Equipment Corporation LSI 11/23 microcomputer and CLSI's complete software package. The micro configured system is less than half the price of a minicomputer-based system. A typical System 23 costs under $90,000. Appleton will automate both its circulation control activities and its public catalog.

Installation of Appleton's System 23 was planned for June 1982 with a temporary installation of a 11/34 computer. The library will initially connect seven terminals, including a CLSI touch terminal which will be used by patrons to access the online catalog.

The bibliographic data base will be developed using Innovative Interface, Inc.'s interface unit to provide an on-line interface to Appleton's OCLC terminal. The library plans to have converted its data base and to begin online catalog and circulation control activities by January, 1983. [Appleton Public Library, Appleton, Wisconsin. Jerry Pennington, Director. (414) 735-6170.]

DataPhase announces appointment

John Beyer has been appointed Director of Maintenance and Support for DataPhase Systems, Inc. DataPhase Vice President of Engineering, Bill Addiletta, said in an interview with LSN that the appointment reflects recognition of the increased size and complexity of DataPhase, which now has 8 people in maintenance alone. DataPhase wanted a maintenance and support director who had defined and specialized skills in the area. Maintenance supervision will be John's sole responsibility, said Adiletta.

The appointment does not reflect a change in Data General's role as the 3rd party hardware maintenance organization for DataPhase Systems.

Libraries with maintenance problems will continue to call the DP Service Desk. If they need to go higher up, John Beyer would be the next step.

The editors believe that the appointment signals a general trend to greater emphasis on maintenance support by turnkey vendors.

Ohio Computer discount plan

Libraries that are considering the purchase of the Virginia Tech Library Systems software will be interested in the following news from Ohio. The State Library Board of Ohio has entered into a purchase agreement with the Hewlett-Packard Company on behalf of Ohio libraries. The benefits of this agreement, the first of what is anticipated will be many such arrangements, will apply to any library in Ohio which orders Hewlett-Packard products. According to this agreement, Ohio libraries will receive a 12 percent discount on all Hewlett-Packard products except for those products classified as "accessories and add ons." Equipment purchases will be based on the terms and conditions of sale as defined in the Hewlett-Packard Federal Government General Services Administration Contract GS-0000-03010.

Any Ohio library wishing to participate in the program will be given a State Library of Ohio contract reference number to be cited on the library's standard purchase order form. Currently only Ohio libraries have this option. However, Hewlett-Packard is discussing similar agreements with other state library agencies.

Word processors rated

Datapro, the New Jersey research firm which specializes in user surveys of office equipment has completed an Honor Roll of Word Processing Systems. A rating was determined for most of the more than 150 word processing systems on the market in 1981 but only the following machines, each rated by more than 50 users, received an overall satisfaction rating of 3.3 or better against a scale of 4.0 points and no single rating lower than 2.7:

Philips Micom 2000/2001	3.6
Lexitron 1202/2303	3.5
NBI 3000	3.5
AS Dick Magna SL*	3.5
CPT 8000/8100	3.4
Lanier "No Problem"	3.4
IBM OS 6/450	3.3

While Wang achieved a high overall rating as a company (3.3), none of its systems made the Datapro Honor Roll.

The survey is scheduled to be repeated in late 1982. Given the dynamic nature of the word processing market, there is every reason to believe that the rankings of the companies will change. Multi terminal system

BRS expands services

Bibliographic Retrieval Services (BRS) has announced extended hours for access to its data bases: 6:00 AM to Midnight Eastern Daylight Time. Lower telecommunications rates will also be available to most users through a new value added network (VAN) known as UNINET. The UNINET rate will be $6 per hour as compared with TELENET/TYMNET charges of $7-9 per hour.

Beginning June 1, 1982 access to the BRS system became available on a non-subscription basis. Users who do not wish to commit to a minimum annual level of usage may access the system for a basic connect rate of $35 per hour plus data base royalties and telecommunications costs. There is an initial fee of $50, which is applicable to any searching done in the first three months.

Among the data bases available through BRS are Agricola, American Men and Women of Science, BIOSIS, Dissertation Abstracts, Management Contents, NTIS, PAtS, Social Science Citation Index, and USBE Most-available Titles. [Contact: BRS, 1200 Route 7, Latham, NY 12110. (518) 783-1161.]

Videotex applications limited

In a recent study, International Resource Development Inc. (IRD) of Norwalk, CT concluded that the market for business videotex will not exceed $500 million a year in the next decade--despite the fact that its technology represents a breakthrough in information retrieval and transaction processing. The study argues that videotex is simply too unsophisticated for most business applications. Many professionals already have access to other forms of information retrieval that are more flexible and powerful. In the second half of the 60s, most companies wanting to establish an on-site retrieval system will be looking for much more than they can get from straight videotex. The trend may be towards integrated data processing/ videotex/management computing systems. Several terminals that can display both videotex and computing formats are already being marketed.

IRD identifies three sections in the videotex market: public service, private systems, and closed user groups. of the three, the last will probably prove to be the most durable, with private system sales never amounting to much and public service all but ceasing to exist by the end of the 1980s.

IRD posits that public systems will disappear as closed user groups evolve. With the projected widespread availability of gateways which enable videotex users to communicate with third party computers various regional and industrial closed users groups will interconnect to form de facto national services. IRD considers that a preplanned national public service such as the British Prestel System, now being widely tested in the U.S., is doomed to failure because it must try to be all things to all people while the nature of videotex demands information tailored to the needs of specific target groups.

NCC goes small

The National Computer Conference went small this year; not in the number of attendees, over 75,000 computer specialists crowded the Astro hall in Houston; not in the number of exhibits, there were more than 700 computer equipment displays; and certainly not in the inside temperature of the exhibit hail, which hovered around 100 F. Small refers to the supermicros which received a large share of the attention of the perspiring masses. The supermicros are the 16 bit machines that will offer the user of a desk top unit support for a much greater variety of applications than is currently possible with the 8 bit micros of which over 1.5 million are now in use.

The IBM 16 bit Personal Computer was lost in the shuffle as companies such as DEC, North Star, Radio Shack, Sony, Vector and others vied for the attention of the critical audience. Only Apple appeared to be holding back. Representatives would make no comment on the Apple IV which is rumored to be under development.

The greatest impact appears to have been made by Digital Equipment Corporation (DEC), the leading minicomputer manufacturer. The company introduced an entire family of micros. The smallest is the Rainbow 100, a personal computer running available CP/M software programs. The Rainbow has two processors (a Z-80 and an Intel 8088) so that it can function as either an 8 or 16 bit machine. The computer "senses" the application and engages the appropriate processor. The minimum configuration is 64KB of primary memory and 2.5MB of secondary storage. The machine can be expanded to 256KB of primary memory and 5MB of Winchester type secondary storage. The base system is priced at $3,245. The next level of machines is the DECmate II, a personal computer aimed at the office environment. There is advanced word processing and accounting capability The DECmate II is based on the PDP 8 and sells for $3,740 for a unit comparable to the Rainbow.

The Professional 300 series consists of two machines, the 325 and the 350. It is designed for multiple concurrent applications. The applications can share a common data base. While not compatible with the PDP/ll mini line, the 300 series offers small system users the sophistication of the PDP/ll series. In fact, it uses the same PDP/II chip that forms the heart of the PDP 11/23 computer. Packaged software will be supplied by 22 vendors who have agreed to rewrite 75 programs for the new machines. No decision has yet been made about CP/M, the operating system that is used by much of the available packaged software. The two models are priced at $3,995 and $4,995.

Radio Shack's new offering-which drew equally large crowds-is a dual- processor, multi-user microcomputer system housed in a variation on Radio Shack's traditional Model II microcomputer cabinet. The machine is based on a Motorola MC68000 CPU, paired with the Model II's Z-60 microprocessor. The MC68000 (which is actually a 32- bit processor) is used for 16-bit operations in the machine. This allows it to address up to 512K of primary memory and use much more sophisticated and user friendly software than can be built into 64KB machines like the TRS- 80 Model II. Unfortunately, the software isn't available yet. That's where the Z-80 microprocessor comes in. It allows all of the Model IIs existing software to be used on the Model 16. The dual processor design carries another advantage for existing Model II owners as Tandy is also selling an upgrade package than turns existing Model II microcomputers into Model 16s.

The basic Model 16 microcomputer comes with 128KB of primary memory and a 1.25 MB disk drive. A second disk drive in the same Model II cabinet (2.5 MB total system storage) adds $800 to the cost of the system, bringing the total to $5,000.

The Model 16 is capable of handling multi-user processing. With the addition of two video display terminals (VDTs), 3 users can share software, data and processing on a single TRS1-80 Model 16. Tandy claims that there is no perceptible loss of performance when the Model 16 is used in this way.

Experts making panel presentations at the NCC estimated that only 20,000 of the 16 bit micros will be shipped in 1982, but expect the volume to grow tenfold within five years. It is the super-micro which is expected to dominate the business education market for systems priced under $100,000. A flood of applications software can be expected within the next year or two.

---

Maintained by Marshall Breeding

Copyright 1994-2024