Proposal: Scalability in Multimedia Architecture

by DR. GREGORY TROPEA, Director Humanities Computing Institute and DR. DENNIS ROTHERMEL, Assistant Professor California State University, Chico Chico, Calif. The past year's conjunction of intense price competition, fast chips, and evolving peripheral and software technology has finally made in-house creation of multimedia applications a realistic possibility for academic software developers. As might be expected, attempts to use these new capabilities bring some previously obscure questions to the fore. Our purpose in this article is to delineate some of the considerations germane to multimedia application development and then to call attention to an emerging design issue that impacts virtually all multimedia applications whose optimal resource file inventory would exceed the capacity of a single CD-ROM. We anticipate that there will be a large number of such applications in the future, perhaps even a majority, once technical feasibility issues are resolved. Two Problems, One Solution Many educators are already involved to some degree in creating multimedia software applications for their own use or for sale. It still takes an appreciable budget, of course, but now we can reckon in terms of four figures rather than five. Beyond budget and time, just about all that's needed to commence multimedia authoring in one's field(s) of interest is a reasonable sense of software aesthetics. Notably missing in this new era of application development is the need for a high level of skill in a general-purpose programming language like Pascal or C. While hardware and software have become sufficiently accessible for academic multimedia projects to succeed without legions of programmers, there are two problems on the horizon that threaten to inhibit multimedia's move from the margins into the mainstream, even if appropriate budget, hardware and authoring systems find their ways into all the right places. The first of these problems is the long-recognized factor of an antiquated publishing infrastructure that tightly restricts dissemination of the images and sounds that are essential to multimedia software. This is a legal/financial problem. The second problem is technological: single CD-ROMs are too limited, in terms of storage space, to create really rich multimedia educational software. While 600-plus megabytes might seems ample room, when it comes to multimedia -- it's not. The good news is that it appears there is a conceptually simple solution that addresses both of these problems at once -- scalability. Educational Need We recognize that the history of technological progress can be read as a story of creating functionality to overcome physical limitations. Before multimedia can function adequately as an educational technology, its applications must be able to represent the imaginations of educators better than they do at present. Classroom experience gives educators at all levels an awareness of student diversity, which translates into a need for flexible software designs. Only if the focus, depth and rhythm of the software are somewhat "adjustable" will a CAI program truly accommodate different strategies of teaching and learning.1 There is much to recommend a design philosophy that could indulge student curiosity with multiple, branching program flows and rich networks of connections among disparate materials. The current technology, centered on the CD-ROM, unquestionably offers dramatic pedagogical improvements over rigidly sequenced displays of jagged, two-color, 40-column images slowly retrieved from 126K floppy disks, but it still is not truly up to the task we have set for it. The presence on the market of sets of CD-ROMs covering a single topic demonstrates that the single-CD-ROM paradigm is already insufficient. One disc just can't physically accommodate designs with such features as large reference databases; numerous or long clips of high-resolution, full-motion video; or an extensive collection of high-fidelity sound segments.2 Even if production costs for the higher-capacity videodiscs become more favorable, it will take a new generation of technology before that format provides practical support for event-driven hypertext tutorials that aspire to imaginal adequacy. The fact is that behind the legitimately exciting multimedia beginnings we are witnessing, is a technology that is still too immature to satisfy the expectations that have been generated. The next step toward the needed maturity is not hard to conceptualize, however, and no difficult technological leap is needed to make it. Move to Distributed Computing For multimedia to realize its potential in education, it must grow out of its closed box and enter the age of distributed computing. What this means in practical terms is that while some sound and image resources can be included on an application's main program CD, an array of supplementary resources would be stored in a technologically and legally accessible format in a different location. Media holding such resources could be as near as inches or as far as miles away from CPUs running the main application or, indeed, from any other program that might access them. All of this could be accomplished using present technology to assemble substantial collections of jukeboxed CD-ROMs or to create heterogeneous networks of CD-ROM and videodisc drives containing terabytes of catalogued sounds and images available to any program that needs them.3 In the near future, an appropriately configured mass storage technology accommodating plug-in, solid-state modules could drive hardware costs dramatically downward for large, networked systems. Given the general direction of change, the need for a distributed multimedia computing standard is not far off. Since it is easy to see these possibilities on the horizon, developers and lab administrators should plan ahead for them. Definition It might seem premature to plan ahead for a standard that hasn't even begun to emerge, but it is not. Certain aspects of any future standard and the system configurations that would support it are logically necessary, and others are very likely. These necessary and likely features of future educational software come together in the concept of a scalable multimedia architecture. The basic idea is that a program delivered on a standard CD-ROM would contain some minimum set of multimedia resources so it could function as a satisfactory stand-alone application, if necessary. However, it would also contain a set of program routines and the required drivers to make use of a pre-defined inventory of other resources if they were available in the operating environment. Under this design, among the needed program routines in any scaled multimedia application would be one that searches for a resource-inventory file. Discussion There are a number of important pedagogical and technological implications of this proposed supplement to the current multimedia architecture. The following sections offer some insights at the main points, which are loosely arranged under general headings; a look at what to avoid is also included. Pedagogical Implications First, since scalable multimedia means that a program could make use of a large array of resources, it is likely that the full range of a program's capabilities would only be available in the near future at institutions, agencies and businesses with a large stake in education and training.4 The hardware and software needed to keep multiple discs online simultaneously would simply be too expensive for most individuals and families, so the pedagogical impact of scaled programs would vary with location. Thus, under this new architecture, stand-alone machines would run basic multimedia programs much like those currently available, while networked computers with extensive peripheral resources would be able to offer far richer experiences.


One design choice, for example, might be to provide still images on the main program disc with text and access software, while full-motion video would be supplied on companion media. Second, the authoring and testing of full-featured, scaled multimedia programs would also require institutional-based resources. One possibility would be the creation of large, online media libraries at regional centers to foster authoring of quality software. Authors might receive grant and technical support for projects whose design and coding were far enough along to benefit from the polishing and resource extension that sophisticated authoring facilities could provide. Technological Implications Third, scaled multimedia programs could be optimized to make good use of true multi-tasking operating systems such as OS/2, Windows NT, Amiga DOS and the UNIX family, including NextStep for Intel, Apple's A/UX, and the coming object-oriented operating systems running on appropriate hardware.5 This is a factor to be considered mostly by providers of multimedia authoring tools; few academic authors will be willing or able to take control of the machine at that level. Adequate performance will most likely entail that the run-time engines of these authoring tools include intelligent caching routines, perhaps with options for application authors who could assign priorities to certain read-ahead operations. Fourth, some newer applications such as program suites and desktop publishing packages, plus others under development, present us with two faces. From one perspective -- when full resources are needed -- they look as though they will be more appropriately run from a CD-ROM than from a hard disk, especially since CD-ROM drive costs and access times continue to plummet. From another perspective, though -- when a subset of the available tools will do and speed considerations outweigh those of cost -- they look as though they will be more appropriately run from a hard disk, at least in the next few years. The need of some users for the high level of functionality enabled by the integration of an extensive set of program resources suggests an increasing frequency of multiple-CD-ROM configurations on individual desktops. If this sort of hardware configuration proliferates, multimedia applications recommending or even requiring dual CD-ROM drives would not be far behind. Fifth, since a variable set of sounds and images would be located apart from the program that is accessing them, there would need to be some way of making their presence known to it. By whatever name, this is a database function. Thus we can expect that host and client (or whatever their analogs would be in other configurations) will need to be aware of a standard data format. This standard could probably be implemented simply as an ASCII file with fixed-length data elements that programs could read and process in any number of ways. If we do not demand fully automatic system configurations with "hot swap" capabilities, setup of this database should be a simple administrative task. There seems to be no need to modify existing device drivers to accommodate the requirements of scalability. Sixth, some interface design and program-logic conventions are needed so that additional resources would not be overlooked when they are available and so that the functionality of design elements could remain consistent across applications. This is another point mainly of concern to authoring system vendors, but since general-purpose programming tools are becoming so much easier to use, any standards should be easily implemented in non-proprietary software components and simple file functions. Commercial Implications Finally, moving a significant part of an application's multimedia resources off the main program disc could simplify royalty and permission matters for developers, reduce distribution costs, plus open up some new possibilities for publishers. Since we can expect that multiple-CD-ROM local desktop installations will appear before long to meet the needs of high-end applications, we can also imagine that some scalable CD-ROM-based packages, with one or more optional supplementary discs containing selected additional resources, would be marketed as well. Individuals and institutions could then purchase the specific resources they need. "Features" to Avoid In addition to features we do want to see, there are some we might wish to avoid. We might not want, for example, essentially non-functional buttons to sit on the screen waiting to display error messages or advertisements for additional resources when pressed. At the program-logic level, a simple list-matching routine might be adequate to notify a program of resources available for inclusion, but only if the resource directory receives subsequent processing. Specifically, some provisions would have to exist in a program's logic to coordinate resources whose use at some point in a program should be contingent upon the availability of other resources (as, for example, in before/after displays utilizing images from different discs). n Reader Response Requested These are some of the key components in the concept of scalable multimedia, which addresses the seemingly disparate problems of legal restrictions on copying of material and inclusion of adequate program resources when storage media capacity is restricted. No technological breakthrough is needed to implement this next generation of multimedia application design, but it will require cooperation among developers and tool vendors, as well as publication of an open, non-proprietary specification. The authors will be discussing this idea with interested parties in education and industry with the aim of developing a workable standard. Comments and suggestions are requested. We'll report on the ideas that emerge and solicit comments as the next step toward creating a standard. Send comments and suggestions to: Dr. Gregory Tropea Department of Philosophy California State University, Chico Chico, CA 95929-0730l e-mail: Gregory Tropea is Critical Thinking Coordinator and Director of the Humanities Computing Institute of the Department of Philosophy at California State University, Chico. Dennis Rothermel is an assistant professor of philosophy at California State University, Chico. Notes: 1.We have experimented with a limited range of such options in our CT_Review program, a text review software package distributed by Mayfield Publishing Co. to adopters of Critical Thinking, by Brooke Moore and Richard Parker (Mountain View, Calif.: Mayfield Publishing Co., 1992). The program's default setting allows students to change screen colors, number of answer choices, order of questions and number of responses before displaying the correct answer, and so on. Copies of CT_Review are available by special arrangement to readers of T.H.E. Journal on request from the authors, courtesy of Mayfield Publishing Co. 2.A CD-ROM will only hold about 500 uncompressed TGA format full-screen (still) images, for example, and that's without any additional access and display software. Full-motion video is becoming more practical, but even with Intel's new Indeo compression scheme that fits a 60-second full-motion clip into 9MB, an application restricted to a single CD-ROM would run out of room quickly. Sound reproduction is a little more promising, but the current practical best of four hours of WAV or VOC files in 30MB, represented at this writing by Vocaltec's CAT, still requires unusual add-on hardware.


3.Because CD-ROM jukeboxes have a lower initial cost than arrays of separate CD-ROM drives, they have an advantage in up-front cost. Since they are mechanical, however, their reliability and maintainability must be factored in. Those who plan for worst cases can see right away that one disabled CD-ROM drive would not affect a system in the same way that as a disabled jukebox. Thus, where real-time reliability is critical, the redundancy criterion favors an array of drives, if everything that's needed will fit on the number of drives that can be added. The issue of network traffic also merits attention. 4.Ronald Kemper, Sr., who administers the Multi-Media Communications Forum, an industry group concerned with a wide range of implementation issues, reports feverish activity aimed at the home market. One of the recognized problems is response time of interactive applications, where even a two-second delay is unacceptable for certain operations. This will limit home-delivered educational applications for the next several years. 5.Without getting into too much detail, we are on the right track if we cut through advertising claims, speculations and wishful thinking focused on the 386-based MPC Level 1 standard to notice that newer operating systems (e.g., Windows NT, NextStep for Intel) are designed for at least a 486 processor. While there may still be some justification for acquiring low-end Macintoshes, no lab should be buying 386 computers at this point; scarcely-tapped capabilities relevant to multimedia applications mean that 486-class machines have a longer future in education than any of the previous chip generations, even given that successor chips to the 486 are already available at reasonable cost.


Further supporting a decision to install 486-class machines is the 1993 MPC Level 2 standard, which calls for a minimum of a 25-MHz 486SX and a double-speed, multisession CD-ROM drive. A better choice would be a faster 486DX processor; DX/2 chips are already widely considered 1994's entry-level component. Dramatically faster video architectures are also a factor adding to the likelihood of extended utility for 486-class computers. For more information on MPC Level 2, see "Multimedia PC Spec Refined" by Tom Quinlan in Infoworld, May 31, 1993, p. 35.

This article originally appeared in the 02/01/1994 issue of THE Journal.

THE News Update

Sign up for our newsletter.

I agree to this site's Privacy Policy.