Web-to-Host Connectivity: Beyond Screen Scraping: Web to Host Meets Thin-Client Computing

While IBM and others struggled unsuccessfully to knock the Wintel alliance off its feet with the network computing concept, their efforts did succeed in solidifying a new set of buzz words to every IT managers’ vocabulary – thin-client computing.

The thin-client market is growing like gangbusters, with a wide range of devices and software available to IT managers seeking to lower total cost of ownership and centralize IT management. Typically, devices, such as IBM’s Network Station, can not only access 3270 and 5250 sessions, but also Windows applications. Microsoft and a bevy of terminal makers promote Windows NT/2000 Terminal Server Edition as a way to keep processing resources on a Windows NT/2000 server. Citrix Technologies adds mainframe, midrange and UNIX servers to this mix.

Ultimately, the future of thin-client computing focuses on the Java-enabled browser, tapping into a rapidly expanding universe of Web-based, mission-critical applications, which also share data with back-end systems. In other words, thin-client computing is fast becoming a form of Web-to-host computing. This means companies need to consider the role of Web-to-host technology in their overall thin-client strategies. Often, these are separate efforts by groups with different agendas.

While Web to host often automatically gets branded as a thin-client solution, the two approaches don’t quite mesh yet. Web to host is typically screen-scraping, either consisting of a Java applet that links directly to the mainframe, or an HTML rendition of a terminal session through a Web server. Thin-client computing usually calls for major re-engineering or development efforts. "Web to host is browser to host, a connectivity project," says Audrey Apfel, analyst with GartnerGroup. "If you’re doing a lot of application development, or putting data into a mid-tier, it’s not Web to host anymore."

In many ways, Web to host doesn’t fit very snugly into the thin-client mold. First, Web to host is not intended to replace fat-client PCs. Web to host evolved from PC-to-host access, which was designed to expand the role of PCs into that of terminal emulators. Efficiencies in Web-to-host computing do not depend on the use of thin-client hardware; any existing computer running a Java-capable browser will do.

Web to host may be well-suited as a terminal replacement strategy, the sweet spot of the thin-client market. However, simply replacing dumb terminals with Web to host will achieve little. The true value proposition of Web to host begins when it extends access to end users that never had terminals.

Nor does Web to host drive application deployment from PCs to a central server, as does a thin-client architecture. Web-to-host computing is designed to access applications that have already been centralized for years.

At this stage, Web to host tends to be deployed for different reasons, and under different circumstances than thin clients. Nevertheless, Web to host offers the same total-cost-of-ownership advantages of thin clients, such as centralized administration, security and software updates. And, as thin-client computing becomes more Web-oriented, and Web to host becomes more middleware-oriented, the line between the two architectures is blurring.

"There’s many shades of gray between what people think of as thin clients and Web to host," says Apfel. "There’s a million definitions. Within an enterprise, people deploy thin-client technology in many flavors."

More Web to host deployments are moving processing to middle-tier Web application servers that sit between users and back-end systems. These servers are increasingly more capable of drawing on enterprise application integration and object request broker-fashioned solutions from multiple back-end servers – as thin clients do.

"Web to host is growing up the food chain from Web publishing to integration," says Apfel. Typical interfaces may pull and present data from applications from multiple servers on different platforms across the enterprise. Code coming out of the mainframe is wrapped, and the mid-tier server includes a database that stores customer information.

On the client side, Web to host may even evolve to include devices other than computers with browsers. Palmtop computers and wireless devices are fair game. In addition, any piece of shop-floor equipment that is powered by Java can conceivably be connected and controlled via a Web-to-host connection.

The convergence of Web to host and thin-client computing is inevitable, as the technologies increasingly overlap. Organizations need to address hardware and software TCO issues within their walls, as well as connectivity issues both inside and outside the enterprise. Not only will a combined thin-client and Web-to-host strategy help drive down costs and increase such connectivity, but it will also prepare the organization for the next wave of innovations in computing.

About the Author: Joseph McKendrick is an independent consultant and author, specializing in technology research and white papers. He can be reached at joemck@aol.com.