For close to twenty years, three intensive efforts have been in progress in the world of technology and business. The Internet industry has been creating technology and deploying operational networks to provide high-performance, pervasive data communications. The Information Systems (IS) industry has been creating, deploying, and refining information technology, centered around databases and now relational databases, for operating massive business machinery using computers.
In the last several years, the IS industry and its customers have discovered networking, and a broad-based push to "client/server" and "distributed database" configurations has occurred. In the same period, the Internet industry has discovered that it is useful to store information as well as move it around, creating the huge foundation of Web technology.
While these activities were happening, a third broad wave of technology evolution has begun in the form of "objects." This technology promises to form a new foundation for programming which is inherently oriented toward distributed multiprocessing. The key for success is to fuse these technologies drawing the strengths and experience from each.
Some have characterized the World Wide Web as the Networking Industry's first attempt at building the world's most primitive object-oriented distributed database system. Despite its limitations, the Web is arguably also the world's most successful such system, if you measure success based on tens of thousands of interacting servers providing information to millions of users.
Similarly, database systems have been very successful, measured by their pervasive use within all forms of corporate computing, and have made much progress in using network technology to create distributed database systems, although on a much smaller scale than the Web has achieved.
Object technology is far newer than either of these, and hasn't yet been deployed in the scale or depth of the Web or database technologies. But it holds much potential for establishing the third technology needed to complement the Web and databases. "The network is the computer" is an old saying by now, but these three technologies will form the three basic building blocks of that computer: the "bus" of the Internet, the "memory" of databases, and the "processor" of object programming.
Technologies succeed and establish themselves as a foundation only when they pass three tests. Functionality is the most obvious; the technology must be able to do what the users want to do. Performance comes shortly thereafter; the technology must perform its work fast enough to keep the users satisfied. The third test is however most crucial - the technology must perform the task with a Cost that is acceptable given what the task involves.
These three tests - F/P/C - dictate whether or not a technology achieves longevity, or whether it blossoms as a fad with great promise but is abandoned as the problems of F/P/C surface.
In the Internet and in the Web, much of the current turmoil derives from the P/C phase that the World Wide Web is living today. Predictions of "meltdown", fears of real-time multimedia traffic, complaints of slow page-loads and overloaded servers and other crises are symptoms of the underlying technology being tested as it becomes more "real" and is widely used.
These symptoms also surface as specific technical initiatives. For example, asynchronous interactions are a technique for dealing with the performance and economic realities of world-wide communications - permitting applications to be deployed that would not be viable if they required synchronous interactions. This permits applications and technology created in a LAN-like situation to remain viable when they deploy to the wide-area environment of the Internet and Web. In the database environment, these techniques appear as a range of mechanisms for distributed operations, from "two-phase commit" through "snapshot replication". These technical approaches are driven, perhaps subconsciously, by a recognition that performance and economics are crucial to a technology as it scales from experimental R&D to become core infrastructure.
These three massive efforts have proceeded largely independently. Most database and IS people can barely spell "Internet". Most "network" people can barely spell SQL. Most "OO" people have not been burned by broadcast storms or tried to manage large Web sites. None of the groups can converse with the others well, being surrounded by almost impenetrable layers of insulating jargon and terminology. Two-phase commit, outer join, referential integrity, symmetric replication? HTTP, HTML, hyperlinks, MIME, ATM, PCS, ISDN, Frame Relay? Inheritance, polymorphism, first-class objects, IDL, ORB? Pick your language...
In the environment of distributed networking and objects, the network itself becomes an object of sorts, and has its own set of characteristics that determine how to use it and how well an application will work when it goes to large-scale wide-area use. Bandwidth and delay are the key characteristics which matter; and the cost of providing appropriate network capacity that is needed for a specific application is the ultimate constraint that determines the success of the technology.
In building applications that use wide-area networks such as the Internet and its in-house corporate clones, design of these applications is critical for determining how they scale and the economics of using them. Application designers make choices, allocating processing, communications, and storage across the network, just as they have made similar choices within a single computer. The next-generation platforms must provide the same kinds of flexibility to make designs that are viable, do scale, and are affordable for widespread use.
The three technologies of the Internet/Web, IS/database, and Objects, each have much to learn from the other. The combination of the three can form the next generation network-computing-application platform. Driving the synergy of these technologies with an understanding of the Functionality/Performance/Cost hurdles is the key.
Jack Haverty