Sun and Oracle caused a stir with their introduction of the network computer (NC) this past spring. Their idea is that the total cost of ownership of today's PCs is too high, opening the door for a low-cost, low-maintenance data appliance. So strong was the reaction of the industry that Microsoft and Intel were forced to quickly pull together a rapid counteroffensive - the NetPC. How does this battle affect the Java developer?
First, let's take a closer look at what the fuss is all about. Sun and Oracle have each come out with their own versions of the network computer. Sun's network computer, the JavaStation, is a thin-client computer based on the 100-Mhz microSPARCª II chip with anywhere from 8MB to 64MB of RAM. It comes with network connectivity and high-resolution video built-in. The JavaStation, however, will have no internal disk drive. The projected price range of the device is $700-$1000. Sun's JavaStation is squarely aimed at the corporate market, with no consumer version announced. The JavaStation will sit on a LAN and download the applications and data it needs on demand. While the initial cost of the device may not be much lower than a desktop computer of similar specs, the long-term cost of maintenance should be lower. Sun estimates this to be about $2500, as opposed to about $12,000 for a traditional PC.
Oracle has taken a different approach to producing a network terminal. It has formed a new subsidiary, Network Computer, Inc. (NCI, http://www.nc.com) to work with hardware and software partners in producing a range of terminals. NCI is working with such hardware manufacturers as RCA, Acorn and others to produce computers for both the consumer and corporate markets. It sees these vendors providing network computer telephones, network computer laptops and many other devices. For example, RCA has announced plans to ship a $300 network terminal for the consumer market in mid-1997.
Microsoft and Intel, on the other hand, have taken their traditional stand on any new attack on the Wintel hegemony. They argue that sticking with Windows will give users the best price/performance ratio because it allows them to maintain their existing set of custom and off-the-shelf applications.
This argument is very similar to the one that Microsoft used against GO Corporation in the early days of the pen-computer wars. As anyone who has read Jerry Kaplan's Startup will know, Microsoft caused no end of trouble for Kaplan's beleaguered venture by questioning the value of GO's drive to create a completely new operating system for the pen-computer - an operating system that would not run Windows desktop applications. Many ISV's chose not to support GO because they could not be guaranteed a large market for their software. Microsoft has recently redoubled its efforts to win the pen-computing space with the introduction of Windows CE. Naturally, a broad range of hardware and software vendors immediately signed up to support the new platform.
The deja vu feeling around this issue was noted even by Paul Maritz, Microsoft Group Vice President: "We have seen this type of scenario before in the PC industry...Our solution is one that builds on the investments a company has already made..." Thus, Microsoft will attempt to convince developers and MIS departments that switching to a new platform will simply result in throwing away all they have built. Microsoft-s argument, however, will not be as easy to make in the NC war since Sun will be bundling in emulators that allow the user to run mainframe screens as well as Windows applications.
Microsoft has even adopted the slogan of the network computer camp in touting a new "Zero Administration Initiative" -- meant to counter Sun-s claim that only with the JavaStation and other NC-s can corporations reduce their overall cost of PC maintenance. The NC initiative, though, has forced the Wintel group to acknowledge what has long been asserted by such data groups as Gartner and Forrester - keeping a desktop PC going is simply too complicated and too expensive.
Where does this leave the Java developer? In particular, if you are planning on developing applications for the corporate market, should you plan on deploying on thousands of JavaStations? Indications are that neither the NC nor the NetPC will replace the desktop computer throughout the corporate computing world anytime soon. There are, however, specific areas where low-cost, thin-client terminals make a lot of sense in the immediate future. Many companies have large numbers of customer service and order entry workers. I have visited call centers with hundreds of operators, each accessing central databases via their PC-s. Since they don-t do anything more than input and retrieve data, the idea of putting a full Pentium with a 1GB hard drive and CD-ROM on every one of these desktops is simply ludicrous. Having some sort of thin-client terminal allows the systems personnel to centrally maintain and update every computer on the LAN.
One difference between the JavaStation and the NetPC is that the Java delivers on the promise of zero administration today while the NetPC group has yet to demonstrate exactly how one achieves low-cost maintenance using traditional Windows applications that are stored locally and not on the server. The JavaStation is meant to be a true toaster-like appliance with few moving parts inside. The NetPC will have a disk drive and other traditional components - making it more difficult to maintain and more complex to administer a security protocol throughout the network.
The good news for Java developers, of course, is that both platforms will support Java. Since the virtual machine (VM) is available on the Windows platform, any application written in Java will run on the NetPC. In fact, indications are that Microsoft-s VM is more robust than others on the market. What will most likely happen is that corporations that already have heterogeneous environments will adopt JavaStations as a more immediate solution for low-cost PC-s. The financial industry, for example, has a mixture of UNIX and Windows software running on its desktops. For them, a JavaStation may be an easier transition. Other companies, however, may remain unconvinced that the JavaStation emulators can really deliver the performance necessary on their existing Windows applications. Like so many trends in the computer industry, this battle will be won on marketing and not on technology.
If JavaStations do take off, there will be increased demand for both packaged and custom Java software. Already we are seeing the number of Java programmers climb to over 200,000. As Java computing penetrates the corporate environment we can expect that number to double. Companies that provide Java-based workgroup solutions, such as Corel, will prosper. JavaStations will also drive increased demand for Java applications servers. These will reside on a workgroup server and maintain a check-in/check-out facility for data and applications. There is much to be done on the server side of the JavaStation equation and we are sure to see new solutions from both traditional developers and startups in the near future.
The bottom line, then, is that Java will most likely prevail whichever way the NC vs. NetPC battle sways. If the JavaStation, though, can gain early market share in the corporate environment, this will drive an exponential demand in packaged and custom Java solutions. Happy coding!