Thin client computing: business savior to the high cost of computing?

Griffin, Ken ; Porter, Harold ; Malley, John 等

INTRODUCTION

Business computing today comes in many different shapes, forms, and flavors. Businesses can rely on mainframe computers, mini computer platforms such as the IBM AS400 platform, Personal Computer (PC) networks, SUN SPARC platforms, or any combination there of. The computing choices available to business today are endless and at many times confusing. Many companies started out developing their systems with focus on a specific platform but then expanded to other platforms to provide specific functionality or cost savings not available on the original platform. Other businesses have allowed branch offices or divisions (especially larger organizations) to select their own platform as long as the functionality required was provided to their customer base, and networking issues were not a problem. Regardless of how most companies have ended up with numerous platforms the fact remains that having several different platforms to support is extremely expensive. Technical support from computing professionals with specific skill sets in each platform is a necessity, maintenance contracts with hardware vendors (most hardware platforms are proprietary, and thus very expensive) is also required in order to provide redundancy and high reliability, and software licensing and contracting to provide operating systems and applications for developers and end users. Add all these factors up and even a large corporation can send itself into bankruptcy if they don't manage their information systems efficiently (Ruley 1997).

TYPES OF NETWORKS

Most computer networks consist of combinations of platforms, operating systems, and applications. A small corporate network might consist of a mini-computer, several PC servers, and up to one hundred PC workstations scattered throughout the company. However, a large corporate network can consist of a mainframe or several mainframe computers, dozens of mini-computers, hundreds of PC servers and literally thousands of PC workstations. The costs of operating and maintaining these large-scale information systems can run into the hundreds of millions of dollars.

With the costs associated with corporate computing continuing to escalate, organizations are looking for ways to reduce costs. In 1995, Oracle CEO Larry Ellison announced that an inexpensive Network Computer (NC) would soon be made available by Oracle that would not only provide an inexpensive hardware solution, but also run an operating system that was platform independent. "The Java VM (Virtual Machine) enables Java-based NC programs to be both operating-system and CPU-architecture independent." This was the first formal announcement of a thin client solution to network computing. With this announcement, many organizations began planning for the implementation of these new workstations with high hopes of reducing corporate computing costs. "The NC is coming, make no mistake about it, and it will be embraced by many managers as the solution to out-of-control costs associated with the conventional desktop computing paradigm. (Molta 1997)" Oracle, IBM and Sun are currently the main supporters of the NC.

A Network Computer terminal or NC is similar to a mainframe dumb terminal. Unlike a PC that can operate independently, a NC cannot function without a server to connect to. "An NC is basically an X terminal (think of it as a PC sans hard drive) enhanced just enough to run a Java virtual machine (VM) (Ruley 1997)."

The NC approach is very attractive to many organizations for several reasons. First, with a workstation that can support any platform, applications can now be run anywhere without being stored and run on the local workstation. "You don't have to be backing up your disk or upgrading your software or worrying about your software getting hit with a virus. All of that will be done for you on a professionally managed network." states Chris Gladwin of Cruise Technologies (Schwartz 1997). Second, with only one type of hardware configuration, corporate support can streamline their operations and support only that one platform. Third, security can be enhanced because the user cannot operate a NC independently. The server has complete control over access to network resources.

Proponents believe that NC's simply won't be accepted. "Very few front-line employees are ready for the roller-coaster ride associated with a radical makeover of their desktop environment, especially if the ticket for all the excitement is a loss of control, performance and flexibility (Molta 1997)". Also, JAVA has not yet proven itself to be reliable and efficient enough for today's high demand business computing. It requires considerable computing resources and is not stable enough for critical applications. Another factor hindering the NC is the fact that JAVA developers and resources are in very high demand, as well as relatively scarce, which is driving up costs. Vern Higberg, VP of IS for Atlanta-based delivery giant United Parcel Service stated, "We're not totally convinced that total cost of ownership of the NC environment is that much better than a well-managed PC environment.(Sweat 1997)" ". NCs process data locally, first downloading JAVA applications from the server, then processing the application, and finally storing data back on the server. After the tasks are completed, the NC deletes the application.

Another Network Computer platform is called the NetPC. The NetPC was introduced by Intel, Microsoft and Compaq as a Network Computer utilizing its own hardware and software but managed by a centralized server. "A NetPC, as defined by Intel and Microsoft, is essentially a Windows-based PC with local computing power and storage but optimized to be managed centrally from a server. (Russel 1997)."" NetPCs offer the ability for the operating system and applications to reside on the NetPC or on the server. If the operating system and applications are on the server, the NetPC downloads both at startup, then deletes them on shutdown. The NetPC is designed to be centrally managed yet still provide the power of standard PCs and greater end user control and security. Also, end users have the choice of storing their data either on the server, or on the NetPC. However, the NetPC does not include floppy drives or CD-ROM drives, so end users cannot install software on their NetPC workstations.

A third category of Network Computers is called "Windows Terminals". This Network Computer is the closest thing to a mainframe dumb terminal currently available. Windows Terminals do no local processing, but simply display graphics downloaded from a server. Windows Terminals offer the highest level of security, can be set up quickly in most cases, and are very inexpensive (since they do no local processing, less powerful hardware can be employed). This platform is also very attractive to large organizations looking to cut information systems overhead, and add additional reliability and security.

The main difference between the platforms is the direct levels of control that both network administrators and end users have over the workstation. Windows Terminals offer the highest levels of security, but the least user flexibility since they are basically dumb terminals, the end user has virtually no control. With an NC, the server maintains almost complete control, whereas with a NetPC, administrators have the ability to allow end users more control over their workstations, since the NetPC workstation can run its own operating system. "The NetPC can be configured as a NC or as a terminal and can alter or change the usage model for that particular device," said Ron Peck, director of network client marketing for Intel's Desktop Products Group.

The main similarity between the platforms is the hardware they use. All three platforms are based upon current PC architecture. Both typically have an Intel or compatible CPU, PC memory, motherboard and chassis, monitor, keyboard and mouse. The reasons for this are obvious: PCs are inexpensive. "The NC will gradually take more building blocks from the PC because it's the cheapest way to build a platform," states Chris Gladwin. However, it must be noted that thin client computers do place high demand on a companies network infrastructure. Because these platforms are so dependent upon network data, they typically require a tremendous amount of bandwidth. This must be considered by any company looking to implement thin client technology with their organization. Additional benefits are that PCs are becoming more and more powerful and supporting enhanced functionality such as multi-media.

NETWORK STANDARDS

As of today, the current Network Computer standard is the Network Computer 1 (NC-1). IBM, Sun Microsystems, Netscape Communications, and Oracle described the NC-1 Reference Profile of minimum requirements for a device called a NC (Gilbertson 1997). According the NC reference file, the NC is not intended to replace PCs but to coexist with them in the workplace. Also, several attributes are expected such as the NC must be architecturally neutral, have a much lower total cost of ownership than personal computers, have a lower entry price than a typical personal computer, and be significantly easier to use and administer (Online 1999). NC-2 is currently being developed and has not been released, but is expected to have additional requirements such as an overall emphasis on promoting network manageability, support for international environments, common cross-platform network booting system and more (Shah 1998).

The theory behind thin clients is a simple one. Provide distributed computing resources throughout the organization that is easily managed, inexpensive, flexible enough to support multiple platforms, and easily upgrade-able. There is a distinct need and market developing for thin client computing even with the PC dropping in price almost daily. "Unlike the PC, the thin client is purely an information-access device, with little of the overhead and few of the headaches of today's PCs.", said Robert Gilbertson, president and CEO of Network Computing Devices Inc.

CONCLUSION

Business computing is attempting to come full circle with its interest in thin client computing. In the 1960's the mainframe computer, with its dumb terminals, was the only productive computing system available. Today, thin client network computing appears to be able to take business computing back towards this arrangement. The thin client ideal promoted over the last few years have, to date, not delivered on many of its promises. Most companies that announced early on that they would be developing thin client workstations have fallen behind schedule and in some cases, not delivered at all. This is unsettling to corporate customers who want to see cost savings as soon as possible. "NCs were initially portrayed as $500 do-everything devices," says Zona analyst Greg Blatnik. "Quite frankly, announced products, technologies, and a whole series of things have not been delivered as predicted." (Hayes 1997).

As with any technology in today's marketplace, these delays can only serve to damage the thin client platform. Other technologies are being produced that offer more functionality, similar costs, and at presently available. These delays are also causing the role of thin clients to narrow. They are primarily being used as terminal replacements in the retail, financial, and manufacturing industries (Hayes 1997).

Will end users accept the thin client platform? Will management reap all the benefits thin client distributed Network Computers have to offer? As with anything in the computer industry, thin client computing could be the wave of the future, or completely forgotten in a matter of months, only time and technology will tell.

REFERENCES

Gilbertson, R. (1998). Future of the PC: NCs: The Fourth Wave, TechWeb, June 23.

Hayes, M. (1997). Reality Sets In For Thin Clients, InformationWeek, October 6.

Molta, D. (1997). Are NCs Really Worth The Price Of Admission? Network Computing, April 14.

Network Computer Reference Profile, Online, www.nc.com.

Peterson, T. D. & Yegyazarian, A. (1998). The New Networked Computers, PC Magazine, May 5.

Ruley, J. D. (1997). PC vs. NC: The Whole Story, Windows Magazine, March 31.

Russel, J. (1997). No End in Sight Over NC/NetPC Debate, Communications Week, August 18.

Schwartz, J. & Marshall, M. (1997). NCs Look More and More Like PCs, Communications Week, March 31.

Shah, R. (1998). Pinning down the ever elusive network computer, NC World, February.

Sweat, J. (1997). IT Executives Divided On Issues Like NCs, Extranets, and Java, InformationWeek, June 9.

Ken Griffin, University of Central Arkansas

Harold Porter, University of Central Arkansas

John Malley, University of Central Arkansas