Database applications developers must focus on hundreds of details every day. It's easy to keep plowing along attending to one critical point after another without taking time to look ahead. Speaking from personal experience, a dizzying array of trends are playing out now, and they have the potential to affect the kinds of applications you build, as well as how you build them.
Here, there, and everywhere. Distributed applications are on the rise. If you are not building one now, you will probably start building one soon. Before the Internet and Intranet explosion, most client/server applications were loners: They did their thing independently of other applications. Sometimes multiple applications would interact at arm's length by working on shared data stored in the same database. Today, several maturing and emerging technologies such as Java, CORBA, IIOP, and DCOM are enabling disparate applications to interact and collaborate across networks, including the Internet, Intranets, and even Extranets. But building distributed applications that work is difficult. There are more pieces involved, and there are more potential points of failure. Issues such as reliability, failover, performance, and monitoring are much more critical. Expect to spend more time planning, designing, and testing.
Turning the tables. Back-end databases are morphing from the banal to the bizarre. Development tools for RDBMS applications have undergone radical changes over the last two decades. Meanwhile, relational databases have been a very stable target: Tables and fields are simple structures, and the datatypes and operators supported rarely change. IBM Corp.'s Relational Extenders for DB2, Informix Software Inc.'s Data Blades, and Oracle Corp.'s Data Cartridges have opened the floodgates for an infusion of user-defined datatypes and the operators that manipulate them. Several dozen third-party companies have announced plans to build a wide variety of extendible datatypes, and some are already shipping. Will your development tool be ready for or bewildered by these newfangled DBMSs? Informix is renovating its New Era development tool to support the Informix-Universal Server, and Oracle recently announced a new Data Cartridge API. But most third-party development tool vendors have not even begun to announce support for extended relational DBMSs and user-defined datatypes. Benefit from this delay by thinking about how your applications and development strategies must move forward.
What goes in must come out. Most of today's application development tools focus on building systems to capture or maintain data. Sure, reporting is always supported, and charts, graphs, and an occasional cross-tab acknowledge that users must also get the data out and make sense of it. Data analysis is largely the province of specialized analytical tools, and many of these tools stress programming-free access and analysis. If they run out of gas, you must find another tool or write a customized analytical application. Many organizations end up building decision-support systems using OLTP-oriented development tools that assume a relational back end. Even though a lot of DSS data is stored in RDBMSs, you can't ignore the growing installed base of OLAP and multidimensional servers. Can your application development tool work with these servers and utilize their strengths so you do not have to write long and clumsy analytical procedures? Few application development tools have set their sights on the data warehouse market, and I find this surprising.
Rethink the role of users. One of our fundamental assumptions is that developers produce applications and users use them. Users generally do not create application components that get bound into the application's executable. Intranet applications are much more fluid and open. Intranet database applications driven by server-side programs, Java applets, or ActiveX controls can link to informational files created and maintained by users. Developers will still guard the glass house enclosing the programmed components of an Intranet, but users can create, contribute, and maintain HTML files that are integrated into Intranets. Users can bang out HTML files by using their word processing program's Save As feature, and these files can be updated without recompiling an application. Unlike earlier client/server and host-based applications that focused almost exclusively on database data, Intranet applications are much more information-rich. When users become cocreators and not just specification providers, development teams will have to adapt to new roles for users.
These four trends merely sample the many forces that are shaping application development now and over the next few years. What other trends are on your mind? Email me with your thoughts.