Database technology has evolved considerably over the last three decades. First of all, relational technology has given us the power to query and manipulate the data in complex ways. Second, object-oriented programming has been widely recognized as a more structured approach to modeling and systems development — it promotes reusability and ease of maintenance, and generally makes software more reliable. The relatively new field of online analytic processing (OLAP) applications has helped us understand a third concept: Multidimensional storage makes applications run faster. Now imagine a database that embraces all three technologies.

InterSystems' Caché 5 is a "post-relational" database that combines all three technologies. It provides both relational and object access to data stored in a multidimensional engine. Additionally, it includes several features that make it well suited for developing Web applications. True to its name, Caché also improves performance by caching the data locally either at the client or on the application server.

Object and Relational Technologies

The database market consists of relational heavyweights such as Oracle, IBM's DB2 and Informix, Microsoft's SQL Server, and Sybase's Adaptive Server Enterprise. On the other hand, object databases such as Objectivity Inc.'s Objectivity/DB, Computer Associates' Jasmine, eXcelon Corp.'s ObjectStore, FastObjects from Poet Software Corp., Versant Developer Suite from Versant Corp., and Database for Objects from db4o have carved a niche for themselves in the marketplace.

The success of object-oriented technology has prompted many traditional relational database vendors — for example, Oracle, DB2, and Informix — to add object/relational features to their databases. InterSystems' Caché stands out from the crowd, because it was designed to provide both relational and object access to the data right since its inception. Moreover, its excellent performance and scalability to thousands of users make it an excellent platform for transaction processing.

Flexible Architecture

Caché's Unified Data Architecture automatically describes data both as objects and tables. Thus, even if you define a database object class, you can still access it using SQL. Similarly, if you create a table using a relational data definition language (DDL) statement, you can immediately access it as an object, too.

Caché Studio provides object access to the database while Caché SQL Manager is its relational counterpart. Additionally, the data is stored in multidimensional arrays, and Caché also provides a direct access path to these arrays. In cases where the object or relational access is not required, developers can access the data directly in the multidimensional format for the fastest performance.

This unique set of features gives the designers the option to model the data in a manner most suited to the type of data involved: as tables, objects, or multidimensional arrays. The Caché Relational Gateway enables Caché applications to access the data stored in traditional relational databases. Caché provides ODBC, JDBC, and Oracle Call Interface access to its database.

Caché installed on my PC in a few minutes and was up and running shortly. The connection and configuration parameters are fairly straightforward and the Caché Configuration Manager provides a well-designed GUI to manage the database.

Caché provides several fail-over features to ensure robustness and minimum downtime. For instance, its Shadow Server can maintain a copy of the primary database that can be used if the main server fails. This device can also be used for nonmission-critical activities, such as reporting without any overhead to the primary database. Another fail-over feature is the use of Database Clusters that include multiple servers accessing the same disk drives. If one server fails, the others can share the load. Caché supports full, incremental, and cumulative incremental backups even while the database is being updated.