Redefining Thick and Thin: Applying the Server-Based Computing Model to Traditional Host Connectivity

Over the past year, thin-client computing has taken center stage in the world of information systems technology. Despite predictions to the contrary, thin-client computing technology is not in trouble today: All indications point to server-based network architectures remaining a hot IT trend.

Over the past year, thin-client computing– a network architecture model in which user applications originate and execute on a server – has taken center stage in the world of information systems technology. There have been significant developments of late. For example, industry analysts estimate that the vast majority of host access will be achieved through Web-based thin-client solutions within the next couple of years. Despite earlier predictions to the contrary, thin-client computing technology is not in trouble today.

All indications point to server-based network architectures remaining a hot IT trend. Driving forces behind the rise of the thin-client computing model include physical and monetary constraints on today’s enterprise networks, decreasing human resources available to support information systems, and an increasing desire to lower total cost of ownership (TCO) – just to name a few. Multiuser Windows NT server platforms, such as TSE and Citrix WinFrame, have increased the popularity of thin-client hardware devices while enabling enterprises to leverage investments in existing PCs. As a result, Citrix founder Ed Iacobucci recently projected that the market for complete thin-client/server-centric solutions (hardware, software and services) will increase to $10 billion by 2001.

Host Connectivity and the Evolution of Thin

Although the notion of thin-client computing certainly does not begin and end with desktop computer hardware, many IT glossaries have defined "thin client" strictly as a synonym for a centrally managed client device, such as a Windows-based terminal, network computer, net PC, or Java computer. Today, however, the concept is more accurately defined in broader terms of a computing model that encompasses an entire category of technologies, software as well as hardware, recognizing one critical objective: server-based delivery of data access to multiple clients that may represent different types of hardware.

Indeed, an increasing number of IS professionals are discovering that the thin-client computing model is applicable to a variety of client hardware and operating systems, including the installed desktop PCs. Centralized deployment and management with a thin-client solution can offer significant advantages, including reduced hardware expenses, lower TCO, simplified network administration, and increased end user productivity.

These benefits can be especially valuable to enterprises seeking advanced solutions for efficient host connectivity: access to legacy data and applications residing on mainframes and other large-scale host systems. Until recently, PC-to-host connectivity has been achieved predominantly through the thick-client computing model, in which terminal emulation software is installed locally on each desktop. This architecture provides the inherent benefits of high performance and flexibility for end users. However, the increasing challenge of reducing costs associated with host access has brought about the rise of thin-client computing systems in which host connectivity software is installed, administered and maintained on a central server. These emerging architectures can be separated into two general types, corresponding to specific IT management objectives and end user computing requirements.

One type is commonly known as Web-to-host connectivity: a thin-client architecture built on Web server-based connectivity software that provides access to host data and applications through a Web browser or standalone interface. A typical example includes local or remote users communicating with a Web server via secured TCP/IP, and establishing host connections through telnet or SNA. Web-to-host connectivity offers the inherent benefits of firm control for the administrator as a result of centralized deployment and management, simplified access for inexperienced and occasional users with the familiar interface of a Web browser, and data encryption security capabilities to protect sensitive information. While Web-to-host connectivity has received significant attention in the IT trade media in recent months, it is not the only form of thin-client computing available to enterprises today.

A second type of thin-client host connectivity, based on multiuser Windows NT server technologies, such as Citrix WinFrame and Microsoft’s TSE, has emerged as a highly efficient method of providing access to legacy data and applications. Multiuser NT solutions apply the thin-client computing model to traditional PC-to-host connectivity, centralizing terminal emulation software on a network server rather than on a Web server.

As with any thin-client computing environment, deploying and managing applications from a multiuser NT server can reduce TCO by streamlining software distribution; centralizing desktop administration; enabling remote control of user sessions; delivering access to a wider range of client types, including both Windows and non-Windows clients, such as Macintosh, UNIX and thin-client devices; and improving performance for remote users. Through the Independent Computing Architecture (ICA) technology provided in Citrix WinFrame and MetaFrame, Citrix/TSE thin-client computing makes it possible to deliver virtually universal access to Windows applications for a diverse array of client devices not supported by traditional PC-to-host connectivity solutions.

In fact, analysts report that approximately half of all Citrix installations are heterogeneous environments: primarily Windows 95/98/NT combined with UNIX, Macintosh, DOS or thin-client devices. A recent opinion poll of visitors to indicated that two-thirds (66.7 percent) of respondents planned to connect at least half of their client devices to multiuser NT servers by the end of 1999. And a survey of Fortune 1000-size organizations by Persoft Inc. found that nearly one-third (29 percent) of multiuser NT respondents currently have heterogeneous computing environments that include thin-client devices, UNIX clients, and/or Macintosh clients.

Perhaps, most importantly, in the era of corporate mergers and enterprise consolidations, a Citrix or TSE platform can easily and cost-effectively extend host access to new groups of users – including those within an organization’s intranet, as well as dial-up remote users and satellite offices. All this while providing capital savings in the form of dollars not spent on new hardware investments to accommodate the seemingly endless compatibility upgrade cycle. Combined with a Citrix or TSE solution, some PC-to-host connectivity software products allow for a cost-effective balance of administrative control over the computing environment (through centralized management of user configurations).

For many enterprises, applying the thin-client computing model to traditional PC-to-host connectivity offers the kind of benefits that demand attention. In the Persoft survey, 63 percent of the respondents plan to or have already implemented a Citrix or TSE multiuser solution. Within that group of organizations, host connectivity ranked second only to desktop productivity suites among the applications to be implemented on the Citrix or TSE platform. When selecting compatible applications, reliability and performance under Citrix or TSE were identified as the two most important criteria.

A Familiar Concept

Multiuser NT implementations such as Citrix and TSE closely resemble the conventional legacy mainframe/"dumb" terminal computing system for multiple-user access. Like mainframe/terminal systems, the thin-client computing model features installation and maintenance of applications on the server. The major difference is that clients connecting to a Citrix or TSE server provide a graphical user interface (the "Windows experience") in place of the proprietary, character-based terminals of the mainframe environment. Of course, Web-to-host connectivity solutions also provide the advantages of centralized administration, an intuitive interface, and the ability to extend legacy data and applications to a wide range of users.

Considering these facts, don’t Citrix/TSE and Web-based host connectivity solutions offer essentially the same thin-client capabilities and benefits? The answer is twofold.

First, it is important to remember that the "look and feel" of host connectivity in a server-based Citrix or TSE environment remains consistent with that of the former thick-client solution. After all, the host connectivity software has not changed; it simply originates and executes from a server and not on individual desktops.

By offering administrators central control over the user flexibility and customization power that feature-rich emulators brought to the mainframe/terminal environment, a Citrix or TSE thin-client host connectivity solution can help reduce support costs and prevent lost productivity resulting from inadvertent misconfiguration by end users. Still, the advanced PC-to-host connectivity interface is not right for everyone. Extranet business partners and inexperienced or occasional users may be better served by the simplicity and ease of use provided by a Web browser interface; Citrix or TSE thin-client host connectivity may be too complex to be practical for these types of users. A simplified interface for easy navigation may be attainable to a much higher degree with a Web browser-based host connectivity solution.

Second, while the Citrix/TSE and Web-to-host versions of thin-client computing both offer implementation and maintenance efficiencies, Citrix and TSE solutions still require installation of either the Citrix ICA protocol or Microsoft’s Remote Desktop Protocol (RDP) on the client device; Web-to-host connectivity solutions do not. Installing these protocols is not necessarily a difficult feat, but it can call for a higher level of end user or IT staff involvement for initial deployment than a Web-based thin-client environment in which a significant number of end users already have a Web browser installed.

Telling Thins Apart

High performance and individual user control over functionality top the list of requirements met by locally installed, thick-client PC-to-host connectivity solutions. To the degree that administrators in Citrix or TSE thin-client environments make the host connectivity software’s capabilities available to specific end user groups, multiuser NT solutions can address these needs. Current Web-to-host connectivity software products, due to the additional server layer between client and host, are often unable to provide the level of performance and individual control that some end users demand. To be at their most productive, power users need the flexibility to customize desktop interface and productivity tools, and this requirement is not met by Web-to-host connectivity solutions that limit control over desktop appearance and functionality. The overwhelming majority of power users, for whom thick-client flexibility is practically a necessity, are found on the enterprise network (within the intranet).

On the other hand, a Web-to-host connectivity solution may be best for computing environments in which administrative control is critical. For example, when providing inexperienced or occasional users (often remote users or extranet clients, such as business partners) with access to legacy data, the added ability to deploy and manage host access from a single point is invaluable to organizations seeking to reduce the cost of maintaining numerous user access points. High-level security via data encryption and other methods, a feature not included in typical thick-client PC-to-host connectivity software products, is another vital requirement met by Web-to-host connectivity solutions for extending host access to users outside the enterprise firewall.

Applying the thin-client computing model opens up new opportunities for information systems managers to improve host connectivity. Not only can organizations realize significant TCO reduction through consolidated maintenance and centralized support for existing end users, but they can also easily extend access to critical legacy data and applications to new groups of local and remote users. In implementing a server-based host connectivity solution, it is crucial to gain a thorough understanding of the various types of users – including their individual requirements and expectations – to whom access will be provided.

About the Author: Kristin Zurovitch is the SmarTerm Product Manager for Persoft Inc. (Madison, Wis.;, which publishes SmarTerm PC-to-host connectivity and Persona Web-to-host connectivity software products.