A Data Approach to Application Integration

As computer systems proliferate in the enterprise, application integration is becoming a frequently heard buzzword. Because of the boom in enterprise resource planning software adoption, plummeting hardware costs, mergers and acquisitions, and the year 2000 problem, companies are being forced to integrate more and more disparate applications, according to Bob Bessin, director of marketing for SmartDB Corp. (Palo Alto, Calif., www.smartdb.com).

Much of the talk about application integration has focused on component-based or process-level integration systems, which tie applications together through programming interfaces. But according to Bessin, there is another solution, called data integration, that often doesn't demand the same intense dedication of resources required by component-level application integration. SmartDB's Workbench product, now available in its fourth version, is designed to help companies integrate data from disparate applications.

Data integration involves many of the same processes as data warehousing: extraction, transport, transformation, validation, mapping and loading. But instead of just feeding data into a data warehouse, SmartDB Workbench is designed to move data between relational databases, enterprise resource planning databases and data warehouses.

Bessin admits that component-level integration is tighter than data integration, making it more appropriate for real-time or complex systems. But he argues that, in most situations, component-level integration can be overkill. According to Bessin, data integration is sufficient for 80 percent of integration projects.

That number may be a bit inflated, according to Jeanine Fournier, senior analyst at the Aberdeen Group (Boston). "I don't necessarily think people are just doing integration at the data level," she says. However, Fournier says that SmartDB's product can also complement process-to-process integration efforts.

Data integration can be useful in several common situations, Bessin says. For example, companies can use SmartDB to maintain consistent reference information, such as part numbers or customer data, across the enterprise. Customers can also use SmartDB to consolidate data, such as financial reports from subsidiaries, into a central financial database. Finally, SmartDB can be used to ensure the results of transactions in one application, such as purchasing, are reflected in financial and other applications.

To integrate data using SmartDB Workbench, customers design templates that map data from one database to another. The template designer is a visual interface that focuses on business rules. The user can view tables and records from the source database, transform the data on a field-by-field basis, and then view the transformed data set before it is transferred to the target database.

SmartDB Workbench 4.0 contains several features that were not available in the previous version, including the ability to monitor and control data integration processes; a refresh loader, which enables automatic recovery from loading failures and incremental data loads on the basis of changes in source data; automation and task management capabilities; improved performance through an in-memory database; and support of Microsoft SQL Server. Version 3.5 only supported Oracle's database.

The Aberdeen Group's Fournier says that the SQL Server support "will certainly open new opportunities to [SmartDB]." She also points out that many ERP vendors are adopting SQL Server, and that these applications are frequently driving integration projects.

A development license for SmartDB Workbench for four concurrent users costs $47,500, with a per-server run-time license costing $7,500. SmartDB also provides prebuilt templates for Oracle applications for $3,000 to $7,000 per template.