Sorting Out Office Automation-A Look at Trends in Local-Area Nets.
Among the sources of confusion are the widely varying categories of automated office systems that have proliferated over the past 10 to 15 years. This article will consider five.
* Information transfer. Electronic message systems transferring information in the form of keyboard characters, facsimile or voice.
* Information retrieval. Computerized assistance in the storage and the retrieval of information.
* Personal processing. Interactive, computer-assisted writing, editing, word-text processing and interactive graphics.
* Conferencing. Telecommunications systems to facilitate multiple simultaneous human interactions.
* Activity management. Automated task-project management, scheduling and ticker files.
In addition to this variety of automated office systems are a number of shades of meaning for the word "local." For the purpose of analysis, three levels of local networks are identified. The device-specific network is a group of identical devices such as word processors, linked together using a communications network specific to the type of desktop devices in use. The work-group network is a work group performing related functions, such as within an engineering department, where multiple desktop devices of various types must be interconnected. The total network is a complete office, office building, or campus environment where many devices of different types must be interconnected.
Until recently, while corporate data managers concentrated their attention on mainframes and minicomputers, personal computers (PCs) made their way into offices and onto shop floors to meet the needs of individual departments. Now that the proliferation of PCs has been recognized and companies have begun to organize their resource, data managers are stuck with the job of linking these PCs into a cohesive network.
Responding to this need are a number of interconnect systems that have appeared on the market over the past few years. One of the more popular of these LANs is Ethernet, but there are many others. Unfortunately, since most of these competing systems are not compatible, there is no standard by which information or resources can be shared between systems of different brands.
To complicate matters further, by the time many organizations become aware of the extent of PC use by their different departments, and attempt to integrate their activities, two or three incompatible LANS may already be in place. Data managers, often brought in after the fact, are faced with the choice of replacing thousands of dollars worth of LAN hardware or finding a means to get their different systems to work together.
Even managers planning new LAN systems have problem to overcome. For all of the attention LAN technology has received, much of the information available is imprecise or conflicting. Worst of all, with the lack of compatibility between competing brands, an incorrect purchase choice can severely limit an organization's options for future expansion.
To help data processing and data communications planners facing these situations, this article will define the issues involved with choosing and implementing local-area networks, and it also will identify the positive strategies that may be followed.
Market for Desktop Systems
Before addressing these issues, it might be helpful to look at the magnitude of the potential market for desktop information-system devices for the years immediately ahead.
Dataquest estimates that there are 58 million desks and workstations in the United States that are potential candidates for some type of desktop communications device. Of those desks, 23.9 million are already equipped with some form of device that is potentially connectable to a transmission system for the movement of data and voice.
Dataquest estimates the breakdown of those devices is 200,000 integrated voice/data workstations; 1.1 million word processors; 3 million electronic typewriters; 10.1 million alphanumeric terminals; and 9.4 million personal computers.
By 1989, the total number of deks will increase to 63.5 million with 50 million potentially connectable devices. The breakdown includes 2.1 million integrated voice/data workstations; 1 million word processors; 6.7 million electronic typewriters; 18.3 million alphanumeric terminals; and approximately 21.8 million personal computers.
Of those 50 million, Dataquest estimates that the following percentage will be connected with some type of desktop information system (communication) device: 93 percent voice/data workstations; 71 percent word processors; 10 percent electronic typewriters; 100 percent alphanumeric terminals; and 76 percent personal computers.
According to these figures, a weighted 76.4 percent of all desktop devices will be connected to an information-sharing system by 1989. The fulfillment of these estimates will entail the addition of 27 million desktop intercommunication devices between 1984 and 1989.
Dataquest estimates that the revenue from the sale of hardware and softwae required to connect these desks to an information-sharing system will average out to $700 per deks. That indicates a market for connection equipment of close to 19 billion dollars between 1984 and 1989.
In addition, the aftermarket for interconnection products necessitated by moves, additions and repairs is estimated to be worth $130 per desk each year. This will result in an aftermarket of almost five billion dollars.
Among those manufacturers vying for a piece of this potentially lucrative market are those who deal in voice/data PBX systems; data-only PBX systems; cable-based local-area networks; modems and data sets; integrated voice/data workstations; word processors; alphanumeric display terminals; office automation systems; personal computer systems; telephone instruments; engineering workstations; and electronic publishing workstations.
In general, every company involved in the data or voice communications industry has a stake in the answer to the LAN question.
As was pointed out at the beginning of this article, there is more at stake in the LAN issue than simply connecting PCs. Different desktop devices have different electronic characteristics that determine how they can be interconnected. More significantly, the processing power available in desktop units is increasing.
This increase in local-processing capacity will continue affect the manner and mode of processing just as it has over the past 25 years. In 1960, the mainframe used batch processing. With the 1970s came the minicomputer and time sharing. As this decade dawned, the microcomputer introduced the beginning of local processing. With the advent of the "nano" computer as powerful as an IBM mainframe of the '70's, the potential exists for almost total local processing.
To best accommodate these trends, a LAN system shuld meet these three basic requirements: First, with most processing performed on a local basis, peripheral resources will be shared. This includes printers, message centers and modem pooling. Secondly, in order to facilitate the most-efficient sharing of information, centralized data bases will have to be created with the capacity for bulk data transfer. Thirdly, electronic messaging and mail transmission will be required to replace the major portion of today's paper flow.
Providing these services will require an effective LAN to display three significant characteristics:
* Local processing and resource sharing requires complete connectivity.
* Centralized data bases and bulk transfer requires high-speed, error-free transmissions.
* Electronic messaging and mail require application-related processing, such as storage-and-forwarding capability, and language translation--essentially the main elements of an intelligent network.
Total connectivity recognizes that a LAN is simply a network inside a network. An effective LAN must provide its users with access to users and facilities inside other LANs, and with the ability to access and use all public networks. Connectivity becomes an issue at the cluster level, at the office-wide level, at the point of connection to wide-area transport, and, finally, within the wide-area backbone distribution.
Integration Requires New Features
Once a LAN is integrated with other ntworks, certain new features become important. These include security, hunt groups, resource eligibility, queuing, local balancing and log-off services. In addition, the complexities inherent in a truly flow-through network-within-a-network create the need for end-to-end network management and control.
While the management and monitoring function may be distributed throughout the network, control should be centralized. In addition, the management and control system should include capabilities for performance monitoring, fault isolation, restoral, configuration and reporting.
The most critical issue in the future of local-area networking is the application-to-application connectivity between end systems. This problem arises from the fact that different vendors use their own specific protocols. The network of the future must provide for open systems interconnection (OSI).
The International Standards Organization, CCITT and European Computer Manufacturers Association are currently working to define the best method for end-system intercommunication. Beginning with an OSI reference model, a seven-level communication architecture was established, and each level of communication architecture was assigned its own protocol. Now specific protocol sets are being defined within each protocol level, and rules are being developed on how an OSI network should be managed.
Given that these open-systems standards are widely accepted, the next step will be the creation of LAN standards consistent with the open-system concept as defined by the IEEE 802 standards committee.
Standing in the way of complete conformity throughout all aspects of all LANs are variables that are highly application-sensitive. These variables include topology, switching techniques, protocols, media and speed.
Topology is the physical configuration of a network. Three basic configuration patterns are currently popular: star, ring, and bus.
A star topology provides for the switching and routing of all data to be performed within a central location. For example, in the voice world, a PBX within a building performs a star-like centralized switching and routing function.
In a ring topology, information flows in a continuous loop until it is extracted at its point of destination. The IBM Cabling System announced in 1984 is an example of a ring topology.
A bus topology broadcats information along a "pipe" to be extracted by the intended receiver. Ethernet is an example of a bus topology.
In the switching area, two alternatives are available: circuit switching and demand switching.
Circuit switching allows for a circuit to be established between two points on an as-required basis. This is accomplished by either providing a physical path (space-division multiplexing) or by creating a virtual path with the characteristics of a physical path (time-division multiplexing).
The demand-switching alternative allows a single path to be shared by all end systems. Path access is on an on-demand basis, yet is subject to availability. Bus and ring topologies use demand switching.
Protocol alternatives are assessed in relation to the chosen network topology and switching technique. The choice among the wide variety of protocols is also determined by the choice of media and speed.
Three generic transmission media--twisted wire, coaxial cable and optical fibers--are available for LAN connections. Twisted wire refers to common telephone wire, universally available in most buildings. Coaxial cable is a shielded cable commonly used with cable TV and is used in the IBM Cabling System. Optical fiber uses light to drive the transmission. Although currently popular in long-haul transmissions, optical fiber is relatively new to LAN applications.
The final element to be considered in planning a local-area network is transmission speed. Speed decisions are closely linked to application. Cost effectiveness can be diminished by a transmission speed that is too low or too high for a given application.
Speed alternatives are: kilobits per second for interactive data; tens of kilobits per second for digitized voice; hundreds of kilobits per second for videoconferencing; Megabits per second for bulk data transfer; and tens of Megabits per second for CPU-to-CPU communcations.
Uses Determine Application
Despite all of these alternatives, the development of a LAN is determined by its application. A look at the general characteristics associated with different configurations is important to provide an idea of the benefits and drawbacks of each.
* Twisted-wire star topology offers low cost; is protocol transparent; has adequate data rate for connects to dumb terminals; can cover a reasonable distance; has internal wiring; provides cnetral control; and is susceptible to interference.
* Twisted-wire bus topology offers low cost; good data rate for file transfer; only covers a limited distance; is non-deterministic--no way of planning line availability; is protocol-sensitive, non-transparent; and is susceptible to interference.
* Coaxial-cable ring topology is high cost; high bandwidth; covers medium distance deterministic--full prior knowledge of line availability; is protocol-sensitive and non-transparent; and has low susceptibility to interference.
What Can Users Expect?
After considering the options available in LANs today, the next question is, what can be expected over the next few years?
Standards have begun to gain acceptance, but it is apparent that there will never be a single set of standards. Even ISO leaves the issue open by providing as many as five protocol alternatives for a single network layer.
It seems clear, and a consensus exists, that the LAN industry can expect extremely high growth over the next five years. Some say that a compounded average growth rate of at least 45 percent is conceivable.
The sheer size of the market will invite a proliferation of smaller vendors to enter some segments of the market. Nonetheless, the vendors who will play the major roles in the LAN market will be those companies that are capable of providing total network solutions.
AT&T and IBM will be major players, are they are the only US companies with the market power to set office-wide LAN standards. It is by no means certain that they will be able to do so. It is possible that open-systems network standards could preempt both IBM and AT&T. In any case, other industry players will closely watch the industry giants.
Since it has been established that there is no single network solution that will solve the entire corporate desk-to-desk connectivity problem, it follows that the requirements of specific applications will be the dterminant factor. Because there are multiple LAN applications, there will be multiple LAN schemes.
The key for companies creating LAN strategies will be to select those LAN market segments that best complement their technical and marketing capabilities. While some may be able to compete across-the-board, it will probably turn out that many will find a rewarding niche in which to concentrate their efforts.
In addition to such strategic decisions, it is imperative that a LAN be recognized as a network within a network. This imposes the requirement of flow-through connectivity that no vendor should ignore. Since the varieties of LAN configurations are unavoidable, a correspondingly wide range of LAN gateways and interfaces will be required.
Lastly, as multi-tiered networks increase in complexity, the necessity for cost-effective and total-network management capability becomes more pressing. The ability to perform monitoring, diagnostics and restoral will remain crucial to effective network management.
|Printer friendly Cite/link Email Feedback|
|Date:||Dec 1, 1985|
|Previous Article:||Interconnected Nets Offer Flexibility and Meet Increased Data Demands.|
|Next Article:||Success of ATM Service depends on Quality Communications to Clients.|