Internet Link Exchange
Member of the Internet Link Exchange

NETWORK COMPUTERS -- THE BIRTH OF A NEW INDUSTRY

By Richard Finkelstein

Part 1


The Internet is about sharing information and related services among millions of people around the world. The nature of the Internet opens up the possibility of redefining the desktop paradigm into one that utilizes more affordable, easier to use, and easier to maintain platforms. Bulky PCs which obsolete all too swiftly can be replaced with a new type of desktop computer, the Network Computer (NC), that is designed specifically for use with the Internet and corporate Intranets.

Network Computers (NCs) mirror our world and our lives. NCs are designed and optimized for sharing knowledge. No person is a community unto herself/himself, and NCs do not expect or require every participant in the network community to be an isolated survivalist. Survivalism is expensive, time-consuming, and subject to rapid obsolescence. People in every-day life do not maintain their own electric, water, and gas supply. As a world - and computing - community we have learned to share specialties and information because that is what makes most economic sense. Sharing allows us to save time and energy and reap the benefits of economies of scale.

NCs can connect into the vast array of services and content called Internets/Intranets (INETs). NCs are also capable of performing common tasks on their own such as electronic mail, word processing, spreadsheets, and graphic presentations. Because of this design perspective, NCs will be much simpler to install, learn, and use than current PC technology. They will have longer life spans, and have a much lower cost of ownership when compared to PCs and traditional client/server technology. Lower costs of ownership and simplicity of operation enable NCs to go where no PCs have gone before. Their impact on the world will be far greater than any computing devices preceding them. What dumb terminals were to mainframes, and PCs were to client/server, NCs are to the INETs.

FROM PERSONAL COMPUTING TO THE INTERNET

Latest surveys estimate that over 20 million individuals are currently using the Internet. Most organizations have either implemented or are planning to implement INET applications. This is an astounding rate of adoption - especially when one considers the short history of the World Wide Web (WWW). While there may be many reasons why organizations are embracing INETs at such an unprecedented rate, in the final analysis INETs are being adopted because they are easy to deploy and to use.

Not only is INET technology cost-effective but there seems to be very little risk. Results are quickly realized and there are already thousands of examples of INET applications which clearly demonstrate the capabilities and scalability of INETs. The beauty about INET technology is that it is not necessary to speculate about possibilities - it is all happening right now. Large and small organizations have implemented catalog ordering, brokerage, reservation, and customer information systems on INETs. These applications are currently serving tens of thousands of people. Some organizations are reporting that their INET applications are being accessed hundreds of thousands of times each day.

It may seem like current INET technology appeared overnight - and indeed, aspects of the INET such as the HTTP (Hypertext Transport Protocol) and HTML (Hypertext Markup Language) are revolutionary - but in fact INETs are the culmination of a trend toward network and server-centricity which began a decade ago.

What started off as a personal computer-based ideology has now evolved into a Web of services and content that is being accessed by customers via desktop browsers. In effect, the PC workstation has moved from a primary role as a content processor and application-specific workstation, to a secondary role as a content access mechanism. The Web servers and INET networks are now dominant. They provide access to vast amounts of content and services. The limited set of applications that can run on an single PC is overwhelmed by the content of INET-based networks and a tremendous richness of functionality.

THE TREND TOWARD SERVER-CENTRICITY

The first generation of PC computing began as a way of economically deploying personal productivity applications (e.g., word processing, spreadsheets) on the desktop. They also gave customers a feeling of empowerment. However, each PC was an island. Data could not be readily shared and there were significant performance, security, and data integrity issues. PCs could not measure up to the reliability of mainframe-based applications.

In the late 1980s, the standalone PC gave way to the concept of client/server processing. In a client/server environment, the relational database management system (RDBMS) is removed from the PC and centralized on a database server. Centralized RDBMSs dramatically improved network performance, data integrity, data security and database manageability, and promoted data sharing. These characteristics were further enhanced with the concept of three-tier application partitioning.

The three-tier model moved some or all of the application from the desktop to the server. Sometimes the application logic was embedded directly in database servers as stored procedures and triggers. In other cases, application code was centralized on "application servers". These servers could reside on their own hardware platform or share a platform with the database server. Centralized application servers and three-tier partitioning further enhanced network performance, application functionality, and data security.

The trend toward thinning clients and more server-centric computing has been dictated by the desire to leverage servers and networks across an ever growing customer community. Servers and networks can deliver more capabilities with increased performance, manageability, and security than individual PCs. Application and database servers can be utilized by thousands of customers and they are easier to manage than distributed PCs since all services and content are located on the server. It is not necessary to disperse and manage applications and databases among thousands of individual PCs.

INETs are a redefinition of this decade-old trend. INETs are made possible by two major technological components: INET networks and Web servers. The INET network is born out of the availability of inexpensive telecommunication services. Interconnected Web servers, which can reside on public or private INETs, create a virtually unlimited framework for information and services. All of these are accessible to millions of customers via standard HTTP communication and HTML display protocols. Participation in the INET community requires very little risk and investment. The simplicity and centricity of the INET model gives it its power.

THE BUSINESS BENEFITS OF INETS

The Customer Perspective

INETs liberate customers from the limitations of client-centric PC computing. Client-centric computing is fundamentally limited by the notion that applications must be installed on each and every desktop PC. As a result, customers are limited to the applications that can be practically deployed and maintained on their individual PC. In some cases portions of the application may be distributed on servers (as in the case of three-tier partitioning), but there is a basic requirement that part or all of each application must be deployed on each customer's PC.

The Microsoft Windows environment enforces the client-centric view of computing. A Windows environment consists of a full complement of operating systems, applications, software and hardware drivers, and Internet software. Practically speaking, an average PC may be equipped to handle at most a dozen or so applications.

The INET changes all of this. Content and services (i.e., applications built from HTML) are fully deployed on networks of servers. There is no need to install any client-centric application software or portion of an application on the client PC. All a client workstation needs is a Web browser to tap into the services on INETs. Thousands of Web-based applications - instead of a dozen - are now available to users. What's more, these applications are current and up- to-date. Customers do not have to worry about getting the latest versions of software nor do they have to worry about getting the latest fixes. The software on the server is the most current and is maintained by the service provider.

Server-centric application deployment does not mean that applications cannot be executed on the desktop. Applications can be executed on client workstations by using Java. Java environments exist on Web servers and within client browsers. Where appropriate, Java applications (applets) can be transmitted from the server to the desktop Java environment. Java applets can be stored on the desktop. This approach is useful when it is necessary to run applications such as word processing or spreadsheets without being connected to an INET. These type of applications can be refreshed from the server when the client reconnects to the INET.

From a customer's perspective, the Web browser environment - the application environment - is very easy to learn and use. Its graphical user interface (GUI) supports document display, input forms and multimedia. The architectural simplicity of browsers makes them highly portable. This means there is a consistent interface to INETs no matter what operating system is on the desktop. Unix Workstations, Macintoshes, Windows, and Network Computers can all be used in conjunction with the INET. In fact, the operating system almost becomes irrelevant. The browser tool and the Java environment define the INET desktop environment.

Because the desktop is no longer dependent upon the underlying operating system, browser software can continue to be useful even as the underlying operating system ages. This helps insulate customers from the extraordinary rate of hardware and software obsolescence that the computer industry has been promulgating for the last decade. Rapid technology obsolescence has an expensive, destabilizing domino effect on an organization's computing environment. Hardware, operating systems, and application software has to be replaced on each and every PC. Network connections that used to work no longer function. Even technical and user knowledge - that has been accumulated over several years - becomes obsolete.

INET browsers are usable as long as they adhere to HTML and HTTP and other standards that are part of the NC reference model. Their useful life is extended because they do not have to be continually upgraded in order to run new types of applications because the applications are all on the Web servers (i.e., Web-centric computing). If an upgrade to browser software is needed the browser is small enough so that it can be download off an INET.

Hardware does not become obsolete nearly as fast as current PCs because the desktop hardware environment doesn't have to support bulky operating systems or never ending growth in application software. The desktop only needs to be powerful enough to run a Web browser and Java applets. Hardware software upgrades are minimized and in many cases eliminated.

In fact, with a reliance on INETs, patches, bug fixes, updates of new applications can all be downloaded automatically without user intervention. Additionally, the only reason to ever upgrade your NC hardware is "performance envy", when faster processors and network infrastructures require hardware upgrades to be realized.

THE CONTENT/SERVICE PROVIDER PERSPECTIVE

With INETs, content/service providers can reach out to thousands or even millions of customers with a minimum amount of effort. Because content and services are installed on the Web server instead of the PC, it not necessary to visit and install applications on the desktop of each potential customer. Organizations thus can avoid costly shipping and help-line support costs that are associated with PC software installations. Upgrading is straightforward. When a new version or an application is installed on the server, it is immediately available to customers.

A further benefit of the NC environment is that it will open a broad new class of markets to content and service providers. The momentum and market opportunity that is bringing millions of new users, each a potential customer, on-line cannot be overestimated. The very nature of INET delivery of applications, all with a consistent user interface, will spur companies to tap the market of users who were traditionally kept out of on-line marketing and sales environments due to the expensive and complicated nature of PCs.

INET applications can be accessed from any desktop that is equipped with an HTML-enabled browser. This makes it much easier for providers who wish to deliver applications to the broadest possible customer community whether the application is deployed on an Internet or Intranet.

Application development is much easier because providers do not have to be concerned with the complex design and deployment considerations associated with supporting multiple client platforms. Furthermore, providers do not have to deal with the testing and debugging problems caused by countless variations of hardware and software configurations that exist in today's customer base.

Maintenance and administration are greatly simplified because all application code is centralized. If there is a bug in an application it only has to be fixed in one place. Users do not have to be notified of bugs, they will automatically get the corrected version of the software the next time they access the Web application. Upgrades can be deployed with similar ease.

ARE PCs A GOOD FIT FOR INETs?

PCs are costly and complex because of their design perspective. They are fundamentally designed as standalone machines that must try to be everything to everybody - all purpose chameleon devices that can be used for games, multimedia delivery, business modeling and analysis, CAD/CAM, and now INET browsing. PCs have to be at the same time simple enough for home users, robust enough for business users, and capable enough to support the thousands of types of applications that potentially have to be installed on them. The tensions caused by these fundamentally disparate design goals creates a very unwieldy environment that is difficult to install and maintain, and which quickly becomes obsolete. In essence, PCs are most useful for "information producers" while NCs are most appropriate for "information consumers".

This problem clearly manifests itself with the latest release of Windows 95. Windows 95 has ballooned into an operating system which requires a minimum disk space of 40 megabytes, and usually upwards to 80 megabytes, while requiring a minimum of 8 megabytes of RAM to run effectively. In fact, this minimum is unrealistic, with many environments finding 16 megabytes of memory inadequate. It is not simply the initial purchasing cost which makes Windows 95 so expensive, but rather the hardware upgrades and time spent installing the system. This process is also extremely destabilizing to the existing system. It is often difficult to estimate the full effects of a new operating system until it is completely installed and in operation.

All applications that have been installed on each PC must be retested for Windows 95, Windows NT, and all new operating systems that may be introduced in the future. Microsoft has already announced that Windows 97 is on the horizon, a prospect that is daunting for a large number of both consumer and corporate PC users. As the number of installed PCs grows, and applications become more and more distributed, it becomes less and less probable that this model can survive. It will ultimately fail under its own weight, complexity and cost.

It is too much to expect that organizations and people can find the resources - time, money, knowledge - necessary to continue this relentless model of client-centric computing which necessarily leads to rapid obsolescence. Not only is this model exhausting, it is no longer necessary. Just as early automobile transportation gave way to shared specialization, so will personal computing. Early automobile owners were expected to know how to fix their own machine. This worked well as long as the industry did not have to scale up and automobiles were reasonably simple machines. This model has given way to specialization and standardization. Now it is more likely that an automobile owner will go to a mechanic for a tune up or major repairs. It is rare that an automobile owner will have the knowledge, much less the inclination, to change their own transmission.

For the same reasons, it is unlikely that consumers will continue to install their own operating systems, hard drives, or memory. Nor is it cost-effective or worthwhile to work through increasingly complex hardware and software compatibility and usability problems with vendors over the telephone.

(Continued)

Part 2

Click Here!