Why Build Data Cartridges?

The Need to Handle Complex Data Objects

Over the years, virtually every industry has evolved sophisticated models to handle complex data objects that make up the essence of their business. By data objects, we mean both the structures that relate different units of information to one another and the operations that are performed on them.

The simple names given these data objects often conceal the complexity of the expertise they embody. For example, the banking industry has many different types of bank accounts. Each bank account has customer demographic information, balance information, transaction information, and rules that embody its behavior (deposit, withdrawal, interest accrual, and so forth).

As will be described below, data cartridges allow you to leverage this expertise by encapsulating this business logic in software components that integrate with the Oracle server. The notion of adding logic to data in a database has been available for some time by way of stored procedures. With the addition of object-relational extensions, the Oracle8i server can now be enhanced by application programmers and independent software vendors to support a new generation of data types, processes, and logic in order to model business objects.

The Need to Operate on Complex and Multimedia Datatypes

At the same time as business models have led to the development of increasingly complex data objects, the revolution in information technology has made it necessary to work with new kinds of data: satellites images, X-rays, animal sounds, seismic vibrations, chemical models -- all these complex and multimedia datatypes are now forms of information that have to be stored and retrieved, queried and analyzed.

Today's web-based applications routinely include many different kinds of complex data. The ability to extend the database to include application-specific data types as well as the business logic associated with these types requires a new class of networked, content-rich, multi-tiered, distributed applications. As will be described below, data cartridges allow you to meet this need by combining scalar and unstructured datatypes in domain-specific components. You can further combine these components to provide both horizontal (across industries) and vertical (niche specific) functionality.

Data Cartridge Domains

The complexity of data objects, which may entail the need to handle specialized data, gives rise to application domains. Put another way: a data cartridge is typically domain-specific. Domains are characterized by content and scope.

In terms of content, a data cartridge can accommodate either scalar data or complex and multimedia forms of data. Scalar data is data that can be modeled using native SQL types such as INTEGER, NUMBER, or CHAR. Complex forms of data include time series, matrixes, temperature and magnetic grids, and compound documents. Multimedia types include video, voice, and image data.

In terms of scope, a data cartridge can have broad horizontal (cross-industry) coverage or it can be specialized for a specific type of business. For example, a data cartridge for general storage and retrieval of textual data is cross-industry in scope, whereas a data cartridge for the storage and retrieval of legal documents for litigation support is industry-specific.

Table 1-1 shows a way of classifying data cartridge domains according to their content (type of data) and scope (cross-industry or industry-specific), with some examples.

Table 1-1 Data Cartridge Domains by Content and Scope

Content	Scope
	Cross-Industry Uses	Industry-Specific Extensions
Scalar Data	Statistical conversion	Financial and Petroleum
Multimedia and Complex Data	Text Image Audio/Video Time Series Spatial	Legal Medical Broadcasting Securities Utilities

Oracle8i furnished the ability to utilize built-in scalar datatypes to construct more complex user-defined types. The Object-Relational Database Management System has now evolved to the point that Oracle 8i now provides foundational cartridges that package multimedia and complex data which can be used as bases for applications across many different industries:

Table 1-2 Oracle Cartridges as Bases for Development

Cartridge	Database Model	Behavior
Time	Ordered list of tuples	Compute Rolling Averages, Compare Time Periods, Construct Calendars...
Text	Tokenized serial byte stream	Display, Compress, Reformat, Index...
Image	Structured large object	Compress, Crop, Scale, Rotate, Reformat...
Spatial	Geometric objects such as points, lines, polygons	Project, Rotate, Transform, Map...
Video	Structured large object of serial (dynamic) image data	Compress, Play, Rewind, Pause...

Following from this, you can see that another way of viewing the relationship of cartridges to domains is to view basic multimedia datatypes as forming a foundation that can be extended in specific ways by specific industries. For example, Table 1-3 shows cartridges that could be built for medical applications:

Table 1-3 Medicine-Specific Extensions to Basic Cartridges

Text	Image	Audio	Video	Time Series	Spatial
Records	MRI	Heartbeat	Teaching	Patient Status	Demographic Analysis

A cartridge providing basic services may be deployed across many industries, as a text cartridge may be utilized within both law and medicine. A cartridge can also leverage domain expertise across an industry, as an image cartridge may provide basic functionality for both X-rays and Sonar within medicine. These cartridges can in turn be further extended for more specialized vertical applications. For instance, any of the cartridges mentioned above could be specialized by being extended by other cartridges:

Table 1-4 Examples of Extensions to a Basic Cartridge

Image
Image -> MRI -> Brain MRI -> Neonatal Brain MRI

In other words, you can develop a cartridge for both horizontal and vertical market penetration.

In summary: data cartridges allow you to define new datatypes and behavior which can then provide, in component form, solution-oriented capabilities previously unavailable in the server. In some cases, where the data cartridge provides new behavior without needing new attributes, the data cartridge may provide PL/SQL packages but not new datatype definitions. Users of data cartridges can freely use the new datatypes in their application to take advantage of the new behavior. For example, after an image data cartridge is installed, you can define a table called Person with a Photo column of type Image