One of the basic trends that has greatly affected reference services in libraries lies in an increased use of CD-ROM-based information sources and less reliance on online database services. When searching an online reference service the method of charging and the difficulty of the user interface require a mediated search where a librarian performs searching on behalf of the library patron. With CD-ROM products, however, libraries often choose to have their patrons use the products directly. CD-ROM products offer a friendlier user interface than do the online database services, and there is no minutely access charge that requires the efficiency of a trained, professional database searcher. Rather, patrons can enter their own search strategies, browse through the results, refine the search, enter new searches and selectively print the results—all without the pressure of online charges.
Basic hardware and software components.
The move toward end-user searching on CD-ROM-based information services means that the library must provide the computer equipment and software necessary for access to their patrons. Access to CD-ROM-based information services involves several components. The computer hardware elements include microcomputer workstations and one or more CD-ROM drives. CD-ROM products consist of both the CD-ROM disc which holds the data and retrieval software. Over 2,000 CD-ROM titles are currently commercially available. Many CD-ROM producers offer multiple titles and can each be searched through the same retrieval software. But each of the major vendors uses different software than its competitors. If a library has CD-ROM titles from different vendors, several different versions of retrieval software will be required. Another piece of software relevant to CD-ROM access on DOS-based workstations is the Microsoft Extensions for CD-ROM. The way that data are encoded on CD-ROM discs differs from the format used with magnetic media, and the 660 megabyte capacity of a CD-ROM disc exceeds the 32 megabyte limit imposed by early versions of DOS. Microsoft developed extensions to the MS-DOS operating system so that DOS applications can access data on CD-ROM discs. The Microsoft Extensions for CD-ROM consist of a device driver, and a program called MSCDEX.EXE. While MSCDEX is a generic program, the device driver is specific to the make and model of CD-ROM drive. Practically all CD-ROM drive manufacturers provide device drivers for their products.
CD-ROM: Models of Access
The number of CD-ROM titles and number of workstations needed to accommodate them will vary from one library to the next. The appropriate configuration of CD-ROM workstations for a library will depend on the volume of use, the number of CD-ROM titles offered, and budgetary constraints. The options available for CD-ROM access in a library generally fall into the following categories:
One CD-ROM database, one workstation. This is the simplest configuration where a single CD-ROM product is mounted on a single microcomputer workstation to which a single CD-ROM drive is attached.
Multiple CD-ROM databases, one workstation. Libraries that offer multiple CD-ROM titles but have only a single workstation for access may choose one of several options. They can have a single CD-ROM drive and have patrons change discs when they move from one database to another. It is also possible to chain multiple CD-ROM drives to a single workstation so that multiple databases can be available without having to switch discs, or to accommodate databases that span multiple discs. CD-ROM minichangers such as the Pioneer DRM-600 offer another option. Although this minichanger can only read one disc at a time, it allows users to automatically switch among six discs.
Multiple CD-ROM databases, multiple workstations. As the volume of CD-ROM use increases, libraries can increase the number of workstations for access. It may be that some CD-ROM databases will be offered on some workstations and some on others, or all the discs can be switched among all the workstations. In this environment, the patron must make sure that the disc for the database that he or she intends to use in loaded in the drive at that workstation. In order for multiple users to access the same database at the same time, it is necessary to purchase multiple copies of the product.
There comes a point when multiple stand-alone CD-ROM workstations become too complex to manage and involve high expense when it comes to purchasing multiple subscriptions to CD-ROM databases. Libraries can then move to a more flexible and more cost-effective solution by implementing a CD-ROM network.
CD-ROM networks. The key feature of a CD-ROM network lies in the sharing of CD-ROM drives among a number of workstations. With the microcomputer workstations interconnected through a network, CD-ROM drives connected to one workstation can be accessed by all the others. This allows a library to purchase a single subscription to a CD-ROM title that can be accessed by several patrons at the same time. Although many CD-ROM vendors charge some additional fee for multi-user network access, these fees generally are less expensive than the purchase of multiple subscriptions.
Components of a CD-ROM Network
Although there are many different ways to create a CD-ROM network, all the approaches share some common elements.
Microcomputer workstations. Even in a networked environment, users access the CD-ROM databases through a microcomputer. For the majority of CD-ROM networks, the microcomputers will be IBM-PC compatible microcomputers running the MS-DOS operating system. Although an increasing number of CD-ROM applications run on Apple Macintosh systems, the great majority of the CD-ROM titles currently available relevant to library use run on MS-DOS systems.
Microcomputers intended for use in a CD-ROM network must be of a certain minimum configuration. CD-ROM applications benefit from a fast microprocessor and a large amount of memory. Among the computer systems currently available, the minimum system that should be considered for this purpose would be one with at least an 80386SX processor with at least 2MB of memory. These systems should be equipped with a 40MB or larger hard drive.
Not only does the retrieval software for many CD-ROM products consume a large amount of memory, but the network software takes an additional allotment of memory. Version 5.0 of MS-DOS includes significant memory management capabilities not available in earlier versions of DOS. Microcomputers on a CD-ROM network should either use this version of DOS, or one of the memory management utilities available such as Quarterdeck’s QEMM.
CD-ROM drives/optical server. Several options are available for connecting CD-ROM drives to a network. The most common approach involves attaching all the CD-ROM drives that are to be made available on the network to a single workstation. This workstation then manages access to CD-ROM data for the entire network, and is called an optical server.
Either self-contained external CD-ROM drives or internal drives can be used on an optical server. If an organization already has a number of external CD-ROM drives, then it is usually possible to connect up to 7 of these to a single workstation, one drive connected to the next in a daisy-chain. The disadvantage of this approach is that the external drives cost more than equivalent internal models and that each drive has a separate power cord. A better approach for networks that require a large number of CD-ROM drives involves installing a number of internal drives in an upright tower-style cabinet. The power supply in the tower would connect to each of the internally-mounted drives. The CD-ROM tower would also include all of the other components of a microcomputer workstation—a motherboard, floppy disk controller, video board, floppy disk drive, and a CD-ROM interface board. It is quite common for CD-ROM optical server towers not to include a hard drive. The software necessary to control the optical server can be loaded from a floppy disk. Optical servers should be dedicated to the single purpose of servicing requests for CD-ROM data. Running other applications on this workstation degrades its ability to perform this function well.
An optical server runs some type of software that facilitates access to the CD-ROM drives throughout the network. The software must be able to notify the workstations on the network about which CD-ROM drives are available, allow the remote stations to logically connect to any of the drives, and access the data on the appropriate disc.
One of the concerns regarding CD-ROM drives generally which especially effects CD-ROM networks regards speed of access. CD-ROM drives are not high-performance storage systems. While the access time of an average hard drive will be less than 32 milliseconds, CD-ROM drives can be as slow as 650 milliseconds. While some of the more recently manufactured CD-ROM drives reduce this access time to around 300 milliseconds, the performance still rates sluggish at best.
To compensate for the inherent slowness of the CD-ROM media, drives that must stand up to multi-user access on a network must use some technique to boost performance. Data caching provides a significant increase in performance. Caching involves storing data read from the CD-ROM in the memory of the optical server. It is often the case that a workstation on the network will access the same blocks of data several times. If the data blocks of data are stored in the cache the first time they are read, subsequent requests for the same data blocks can be services almost instantaneously. Although the chances of the requested data being available in the cache are relatively small, even a small percentage of cache retrievals versus new disc reads can lead to substantial performance benefits. Some optical server control programs implement a read-ahead strategy where the system anticipates data requests by looking beyond the current data request and placing it in the cache, thus greatly increasing the chance that the next data request will already be available.
Several vendors offer software that performs this function on an optical server. The main offerings include: CD-Connection from CBIS, CD-Net from Meridian Data, and OPTI-NET from Online Systems, Inc.:
OPTI-NET Version 2.0 for MS-DOS
Online Computer Systems, Inc.
20251 Century Boulevard
Germantown, Maryland 20874 (301) 428-3700
FAX (301) 428 - 2903
CD Connection/CD Server
5875 Peachtree Industrial Boulevard, 100/160
Norcross, Georgia 30092
(404) 446 - 1332
(404) 446 - 9164 (fax)
(404) 446 - 0485 (bbs)
(404) 446 - 3337 (technical support)
CD Net Version 4.2
Meridian Data, Inc
5616 Scotts Valley Drive
Scotts Valley, CA 95066
(408) 438 - 3100
(408) 438 - 6816 (fax)
Network cabling. Another major component of the CD-ROM network consists of the cabling that interconnects the workstations and servers. The style of cable used depends on the network operating system and the type of network interface cards selected for the network. Several cable types prevail among the available options.
The two most popular types of networks include TokenRing and Ethernet. Either of these network types will support the loads imposed by a CD-ROM network if properly implemented. Both are well defined, highly supported and widely implemented. Standards exist for each type: Ethernet by IEEE 802.3 while IEEE 802.5 defines TokenRing. As long as they are adhered to, these standards ensure that products from all the vendors will be compatible with each other.
These two main network types differ in several respects. Ethernet moves data on the network at 10 megabits per second and TokenRing networks come in both 4 megabit per second and 16 megabit per second varieties. The low-level communications rules, called network protocols, differ significantly between these two approaches. Both Ethernet and TokenRing offer quite good network performance and are both supported by a large variety of hardware and software products.
Ethernet networks can run on several different cabling types, but the two most common are called thin-wire Ethernet and UTP Ethernet. Thin-wire Ethernet consists of a RG-58 coaxial cable that connects one workstation to the next in a daisy-chain configuration. UTP, (unshielded twisted pair, also know as 10BaseT Ethernet, uses basically the same type of cable as telephone systems. Instead of the chained configuration, this type of cabling requires a discrete cable between each workstation and a central hub. While this strategy may involve a greater amount of cabling, the resulting network is more tolerant of faults in that a broken or disconnected cable disrupts only a single workstation instead of the whole network.
TokenRing networks employ shielded twisted pair cabling. The shielding of the cable protects the data from disruption from external sources of interference. TokenRing cabling, like UTP Ethernet, follows a star configuration where each workstation connects to a central network hub with its own discrete cable.
Network adapter cards. Each workstation or server on the network needs some adapter card that connects it to the cabling system. The network interface card mounts inside the microcomputer’s system unit in one of the bus slots, and has a connector to which the network cable can connect. The type of network interface card to be used depends on the network cabling and the network software selected. Not only must the network implementor select between general categories such as TokenRing or Ethernet, but also among the dozens of brands and models of boards within each category. Again, it is important to make sure that the network interface card, network cabling system, and the network operating software are all compatible.
Network operating system. The network operating system controls all the activity relating to sharing resources among the stations on the network. It establishes communications among the network stations and moves information along the cabling system.
The network software facilitates the sharing of resources among a group of computer systems. The resources most commonly shared on a network include disk storage and printers. Most networks will include a file server which services the disk storage needs throughout the network. Each user of the network will be allocated an appropriate amount of storage space on the file server which can be accessed just as if it were connected to their local system. Network printers work in a similar way. Many networks will include a shared laser printer to which any authorized user may access. The network operating system takes care of re-routing requests for shared resources on the network from the individual workstations.
Scores of network operating systems can be found on the market today. Most of the popular network operating system work well with CD-ROM networks.
Novell NetWare currently dominates as the network operating system of choice for medium-sized to large networks. NetWare runs on a dedicated file server, and employs sophisticated server software to provide high-performance access to data on the file server. Shared printers can either be attached to the file server or to a workstation located remotely on the network. NetWare uses a network protocol called IPX to transport information about the cabling system.
Other network operating systems can also support a CD-ROM network. A number of these are based on a network protocol called NetBIOS. These include the Microsoft LAN Manager, PathWorks from Digital Equipment Corporation, Artisoft’s LANtastic, the IBM’s PC-LAN, and others.
In a CD-ROM network, the most common arrangement involves the use of a general-purpose network operating system to provide network communications and access to the CD-ROM retrieval software on a central file server. None of the high-performance network operating systems support CD-ROM directly. CD-ROM drives can be added to these systems through the optical server approach as described above. The optical server control software cooperates with the network operating system to provide shared CD-ROM access.
Another approach for adding CD-ROM access to a Novell NetWare network that has recently gained popularity is one where the drives can be connected directly to the NetWare file server instead of to a separate optical server. Although NetWare does not support this configuration natively, third-party software products such as SCSI Express form Micro Design International and Corel Systems’ CorelDRIVER allow CD-ROM drives to be connected and mounted as a NetWare volume just as would a magnetic drive. The general advantage to this approach lies in the ability to take advantage of the powerful capabilities of the NetWare server. In most cases, this server has a very fast processor with a large amount of memory for data caching. The caching facilities available allow for a higher level of performance from a CD-ROM drive than would be possible on most optical servers.
Micro Design International, Inc
6985 University Boulevard
Winter Park, FL 32792
(407) 677 - 8333
(407) 677 - 8365 (fax)
CorelDRIVER Version 1.10
Corel Systems Corporation
The Corel Building
1600 Carling Avenue
Ottawa, Ontario, Canada K1Z 8R7
(613) 728 - 8200
(613) 728 - 9790 (fax)
CD-ROM networks have proven themselves to be highly successful for libraries. This technology began to emerge in libraries beginning in about 1985 and has quickly proliferated. Libraries can be confident that the products currently available for establishing a CD-ROM network will work with a high degree of reliability. While five years ago it may have been difficult to put together a CD-ROM network that would work at all, the problem today is selecting from among a number of competing products that are all known to work well, albeit with varying features and performance.
CD-ROM networks, as with most other technologies, are subject to the phenomenon of rising expectations. A network that serves its intended function well will be inevitably be called on to do more. As library patrons experience the benefits of access to CD-ROM from within the library, they will eventually request access from the computers in their own offices. The trend, especially among academic libraries, is to provide access to CD-ROM-based information services not only from within the confines of the library, but to expand access through campus-wide networks. Even if a library implements a CD-ROM network with the initial purpose of serving patrons from within the library, allowances should be make for future expansion of access. Choices of network products should be made in consideration with their ability to integrate with any other networks that exist, or are planned, in the corporate or campus environment.
Betts, Kelly, Pastrick, Greg and Perratore, Ed. "Networking CD-ROMs: the Power of Shared Access." PC Magazine 10(22) (December 31, 1991).
Breeding, Marshall, ed. 1992. Library LANs: Case Studies in Practice and Application. Westport, Connecticut: Meckler.
Desmarias, Norman, ed. 1989. CD-ROM Local Area Networks: A User’s Guide. Westport, Connecticut: Meckler.
Derfler, Frank J., Jr. 1991. PC Magazine Guide to Connectivity. Emoryville, California: Ziff-Davis Press.
Hancock B. 1988. Designing and Implementing Ethernet Networks. Wellesley, MA: QED Information Sciences.