On Air: What Goes Around...The Business of Web-to-Host
Author’s note: In the April issue of Enterprise System Journal, I presented a high-level overview of the trends and enabling technologies driving the fledgling Web-to-host market entitled On the Bleeding Edge of Web-to-Host. In this follow-up article, my discussion will focus on the business of Web-to-host, the social aspects of Web-to-host technology, its impact on desktop computing, and the development platforms being used.
In the previous article, I discussed IDC’s highly optimistic forecast for Web-to-host revenue, which they expect to surpass a billion dollars by 2001. However, it appears that the market may not be developing quite that rapidly, primarily because of the robust installed base of fat client technologies, and the giant leap that is required to adopt Web-to-host solutions. A confirmation of this contrary view may be garnered from the conspicuous lack of advertising for Web-to-host products. Generally speaking, if a technology segment generates more than $50 million in revenue (which is what IDC forecast for this year), it’s almost a given that some companies will advertise, but one can page through magazine after magazine without encountering a single ad for this technology.
In order to get a feel for the pace of Web-host adoption, we did a quick survey of customers who attended a recent seminar on the subject. Of 60 people invited to the survey, only 24 showed up. Of those 24, only 15 responded to our survey. Of the 15 completing the survey, only one indicated that a pilot study was underway, three had software in-house and 12 were there for general interest or were just gathering information. The results, which are fairly well qualified, indicate that the Web-to-host market is not developing at the breakneck speed forecast by our market research brethren. (Remember that most market researchers consider 50 random surveys to be adequate for market analysis and extrapolation.)
In an upcoming Gartner Group Report entitled, Strategic Analysis Report on Terminal Emulation and the Evolution Towards Web Technologies, a Gartner Group analyst lays out the playing field of Web-to-host vendors in the dreaded "magic quadrant." Although we will not reproduce the exact position of the players in their respective quadrants, the companies mentioned are listed below. Interesting, IDC’s market leader OpenConnect is positioned in the visionary segment of the quadrant (see Table 1).
Web-to Host Challengers Web-to Host Leaders Web-to-Host Niche
Hummingbird Attachmate Eicon
NetManage IBM Futuresoft
Esker Wall Data Persoft
Table 1: Gartner Group’s Magic Quadrant Players
A Socialist Form of Computing
Web-to-host technologies and the distributed object-oriented computing paradigm will ultimately transport us back to the timeshare computing days of the 1970s. Web-to-host technology represents an important cultural change in personal computing. Some might consider this new paradigm a socialist form of computing. Personal computer users have evolved to be individualists and are accustomed to computing autonomy. Loss of control over their computing environment (i.e., the instant availability of their familiar and powerful personal productivity applications) may have a negative impact on PC users’ productivity, as well as their perception of and willingness to cooperate with MIS.
The social and cultural aspects of computing are often overlooked, and generally revolve around color definition and user interface design. There is nothing futuristic about Web-to-host technologies, and hidden in the hyperbole and excitement of the technology is a clearly defined Orwellian objective: take back control of the desktop. Will Web-to-host users become second class computing citizens? Will the first generation of Web-to-host solutions feature rich high-performance applications to the desktop? The answers to these questions are "yes" and "no." Web-to-host technology will create a second class of computing users that will interact with their computing environment through the lens of a browser. For it is extremely unlikely that the first generation of Web-to-host solutions will deliver anywhere near the functionality and performance delivered by fat client technology today.
Performance Is King
One of the most defining characteristics of any desktop application is its performance, and performance is simply perceived as how fast the application responds to a user. Performance is judged by the user in relation to their entire computing environment, and in today’s computing environment many fat client TCP/IP applications, such as terminal emulators appear to respond nearly as fast as the applications users regularly interact with on the local hard drive. Implementing Web-host terminal emulation may indeed send users back to the future as application performance is degraded by the immaturity of the technology, and its dependence on a network-based Web or application server.
Performance, however, will be the key to selling Web-to-host technology solutions to the users. And in our quick-and-dirty seminar survey, 13 of 15 respondents indicated that performance was the number one feature importance when considering Web-to-host. Many Web-to-host vendors have apparently realized this and are hedging their bets, providing a smorgasbord of connectivity options, such as Java applets, ActiveX controls and HTML gateways. This is also a strong indication of fledgling market.
In the case of Java-based Web-to-host solutions, desktop application performance may be impacted by a host of new factors including applet size, the number of applets (depending on initial applet functionality), applet caching, origination of the applet, gateway load, etc., not to mention the host load. And like all other network-dependent applications, bandwidth can certainly play a major role.
Web-to-host application performance will be evaluated on how quickly the applet/ActiveX control/HTML gateway responds to the user, and how quickly they can be connected to and interact with host information and applications. In this case user perception is unavoidable reality.
Interestingly, in the course of our first article we overlooked one of the most significant enabling platforms for Web-to-host software: the Windows NT server. Windows NT server is already playing a major role as a development platform for Web server/gateway solutions, and as CORBA-based enterprise application servers. If Moore’s law holds true its only a matter of time before inexpensive Intel-based NT servers bulked up with memory and enormous processing power become ubiquitous.
However, although the price performance characteristics of Intel/NT servers are extremely attractive, enterprise strength scalability is the hallmark of the UNIX platform. Behind the smoke and mirrors (i.e., marketing), Windows NT was not developed as a multi-user operating system, and even today doesn’t deliver the scalability and reliability of UNIX. This situation has spawned a great deal of controversy and debate; however, when you have to change 200-plus kernel level files in an operating system to make it multi-user, (a la Citrix), I trust most IT managers would agree it is not multi-user in the UNIX sense of the terminology. I have often written about the Windows NT tsunami and have been a big proponent of NT, but on my most recent tour of Wall Street, I asked several IT managers if NT had made it out of the print and file server ghetto yet. The answer was "no."
Enterprise-strength scalability will be key in Web-to-host solutions, and the CORBA/UNIX platform is poised to deliver the highest level scalability. So why aren’t more vendors developing Web-to-host solutions on Sun, HP or IBM UNIX? This is indeed an interesting question, especially considering the evolution of CORBA as an open standard platform for distributed object-based computing.
Of course, this leads us into the next major debate of the new millennium: CORBA vs. DCOM. Although this may be a debate many perceive to be raging throughout the industry, many customers, when questioned about their preference for an object model, merely state that it’s not an issue. This may indicate that they are not ready for object-oriented computing, or that they will simply wait for it to work itself out. One thing is for sure, in the enterprise multi-vendor computing environment of today, the days of proprietary computing are over and interoperability is the mantra.
ABOUT THE AUTHOR:
Peter J. Auditore is Vice President U.S. Marketing at Hummingbird Communications Inc. (Mountain View, Calif.). He can be reached via e-mail at email@example.com.
On Air is an associated forum for guest viewpoints on the latest technology trends and issues. If you want to be heard On Air, send your topic proposal to the Editor in Chief Charlie Simpson at firstname.lastname@example.org.