The History of Universal Data Access

All commercial database management systems (DBMSs), such as IBM DB2, Microsoft SQL Server, and Oracle, include a proprietary or native API for accessing the data they contain, as well as a wire protocol that describes how they transmit data over a network. For example, the native API for SQL Server is known as DB-Library (DBLIB) and the wire protocol is called Tabular Data Stream (TDS). The most direct way to build an application that calls any of these DBMSs is to use the native API included with the database.

However, this approach has several limitations. First, it makes combining data from multiple DBMSs difficult because developers must deal with different APIs for each product-for example, if a table of customer information is stored in DB2 and the orders table in Oracle, an order tracking application becomes quite difficult to develop. Second, data might be stored in a nontraditional data store (ranging from large mainframe databases to simple Excel spreadsheets or text-based files) for which no standard data manipulation API exists. Without some intermediary service, developers would be forced to access the file directly to extract information, making it difficult to build applications that combine data from multiple sources or to share code between applications.