ILS 603
Prof. Howard Besser - Image Databases

Imaging Across the Network- Issues and Trends

Susan Hinton
October 13, 1995

"IMAGING -- Recording 'human-readable' images - pictures, images, motion, text, etc. - into 'machine-readable' formats - microfilm, computer data, videotape, OCR output, ASCII code, etc."

(Andy Moore, Moore's Imaging Dictionary)

"Software is the limiting factor as speed improves. The software today stinks. It isn't close to the vision of what we can do. Software will get better."

(Jim Allchin, Senior VP, Business Systems Division, Microsoft Corp.)

"Clarke's third Law: Any sufficiently advanced technology is indistinguishable from magic."

(Arthur C. Clarke)


Today, companies are image-enabling their applications in any number of ways, to add value to their business practices. For instance, a customer service department attempts to manage their paper glut by deploying a workflow system which first scans and digitizes, then annotates and processes, incoming orders and correspondence; the training department updates their on-line manuals by incorporating Apple QuickTime clips which demonstrate equipment operating techniques; and the human resources department links the personnel file to an employee photo database to improve their security procedures. Each of these simple examples yields a dynamic, electronic, compound document - a container which holds pointers to a variety of object types. This paper attempts to explain why these documents need to be managed differently then text files from an access/control perspective and also why they must be given due consideration when designing networked solutions for storage and retrieval of these typically large files.

Imaging Goes Mainstream

There are several indications that imaging is poised to make a pronounced impact on business applications. BIS Strategic Decisions Inc., a market research firm, has followed the increase in flatbed scanner purchases from 300,000 units in 1993 to a predicted 580,000 by the end of 1995, with the purchasing trend shifting from graphic artists to corporate desktops. They are predicting this growth will continue and projecting 850,000 units will ship in 1997 and approximately 1.2 million by 2000 (Connor S8).

Software alliances between Microsoft and Wang for the inclusion of Wang imaging technology in Windows 95 (by the end of 1995) will bring imaging technology to millions of new users. Additionally, Lotus Development Corp. is including free imaging software with its Lotus Notes and cc:Mail products. The Lotus Image Viewer permits enhancement, adjustment, printing, faxing, and mailing of images and will display any image file or fax in the common formats (BMP, PCX, Photo CD, LNDI, JPEG, GIF, RTF, TIF) (Ginsburg S20-S21).

Systems and network administrators need to plan for coming onslaught of desktop imaging applications. How should these compound documents be filed? Where will they originate? What are their destinations? How will users interact with the data? These are just a few of the many questions that need to be addressed, as the requirements for storage and bandwidth increase (Ginsburg S22).

Documents and Databases

Document management systems (DMS) that integrate workflow, imaging technologies, and storage management, are emerging as standards-based applications capable of providing controlled access, organization, location, and retrieval capabilities for compound documents across corporate LANs and WANs. Driven by user demand for common user interfaces across the numerous vendor applications, the Document Management Alliance (DMA), composed of all the major software developers (Novell, IBM, Microsoft, Hewlett-Packard, etc.), released its interface specifications in July. DMA's standardization efforts focus on three elements: "a common interface, a uniform applications programming interface and an object-based data model for standardizing access to library services" (Ginsburg S10).

Currently, most corporations are running their document management product on top of a relational database (RDBMS). RDBMSs are well-suited to handle the cataloging and indexing type functions that are required by a DMS to store and retrieve documents and their associated profiles and meta-information. Storage requirements for such databases must be analysed for capacity (gigabytes vs terabytes) and user-accessibility: fast access, networked disk drives for on-line data; slower access for the infrequently used near-line data; and off-line for archived data which could require human intervention to retrieve (Ginsburg S24).

As documents grow in complexity (embedded with reusable multimedia parts) a trend towards object-oriented databases is anticipated (Rogers S29). Corporations are expected to grow repositories for shared objects such as sound, image, and video and manage them across an enterprise network -- to further enhance activities such as collaborative research projects, employee distance learning programs, and multimedia customer presentations. Document management systems that can sufficiently organize documents at this level of complexity do not currently exist. However, the goals of such a DMS appears achievable within the coming decade .

Network Evolution

Speculations on the future roles of integrated document management systems in the corporate environment need to include the projected evolution of network communications. There can be no doubt that the network has become, in the words of Novell's CEO, Robert Frankenberg, "the central nervous system for most companies" (Haight 19). But it is a nervous system that is required to expand to accept greater capacity -- desktop videoconferencing, groupware, interactive multimedia, voice and image embedded e-mail, hypertext -- and connections - telecommuters, mobile users, suppliers, partners, customers.

The demands of running image-enabled applications could severely challenge or even cripple networks with insufficient bandwidth. A compound document could easily contain many megabytes of data, even after hardware and software compression. Consider the simultaneous transmission of such files over an Ethernet LAN with a shared bandwidth of 10 Mbps. The LAN would easily be saturated by the traffic. The situation would be worse if the transmission occurred over a low speed (56 Kbps - 1.544 Mbps) WAN connection. Network administrators need to clearly understand the bandwidth demands these applications place on their existing infrastructure and perform capacity analysis measurements to expose any limitations of their networks.

One technology that is frequently cited by information industry analysts as a way to bring networked multimedia to the desktop, is ATM (Asynchronous Transfer Mode). The ATM Forum was initiated in 1991 by a group of four computer and telecommunication vendors and has rapidly grown to over 500 active members. It was organized to accelerate the definition of ATM technology and promote communication between equipment manufacturers and end users. The Forum is not a standards body but rather, a consortium of companies that writes specifications for ATM products which are then passed up to the ITU-T standards body. According to the ATM Forum, the following concisely lists the benefits of ATM technology:

  1. One Network - ATM will provide a single network for all traffic types - voice, data, video. ATM allows for the integration of networks improving efficiency and manageability.
  2. Enables new applications - Due to its high speed and the integration of traffic types, ATM will enable the creation and expansion of new applications such as multimedia to the desktop.
  3. Compatibility - Because ATM is not based on a specific type of physical transport, it is compatible with currently deployed physical networks. ATM can be transported over twisted pair, coax and fiber optics.
  4. Incremental Migration - Efforts within the standards organizations and the ATM Forum continue to assure that embedded networks will be able to gain the benefits of ATM incrementally - upgrading portions of the network based on new application requirements and business needs.
  5. Simplified Network Management - ATM is evolving into a standard technology for local, campus/backbone and public and private wide area services. This uniformity is intended to simplify network management by using the same technology for all levels of the network.
  6. Long Architectural Lifetime - The information systems and telecommunications industries are focusing and standardizing on ATM. ATM has been designed from the onset to be scalable and flexible in:
ATM is a switched service which uses a fixed length (53 byte) cell-relay transmission with guaranteed switching speeds ranging from 1.544 Mbps to 622 Mbps, depending on class of service; it moves bits of information, regardless of their data content in an efficient manner and is not tied to a specific type of physical transport. ATM is usually discussed in the 155-622 Mbps range and can achieve these speeds because it does not provide error checking or correction services (Minoli 115). It depends on the digital carrier systems to provide relatively error-free transmissions and relies on upper layer protocols to check for errors (American Research Group 1-20).

ATM equipment has already been implemented at many corporations and universities. Others are selectively testing vendor offerings at the LAN and WAN levels. Due to the lack of maturity of the technology, associated costs of equipment and services are still high compared to FDDI and Ethernet. Additionally, there are current and emerging standards to carefully examine when considering vendor equipment, to integrate into an existing network. Proponents hope that over time, the line between LANs and WANs will converge on ATM, providing a simple and efficient standards-based network, designed from the beginning to accommodate the simultaneous transmission of data, voice, image, and video.


The trends discussed in information content, organization, and access of compound documents, present many challenges to corporations. No 'plug and play' solutions exist, however software companies have offered some standards which will lead to further interoperability between applications. Additionally, network technologies exist which can scale-up to drive these applications from the desktop.

Bibliography and Suggested Readings