The simplest architecture for an Omnidex application is a single server that houses an application, a database and Omnidex. Optionally, web services may be active on this server as well. In this model, applications connect to Omnidex using ODBC or JDBC, and Omnidex interacts with the underlying database as needed. The application can be a third-party business analytics tool, or it can be a custom program serving pages to a web server. Omnidex utilities such as the Omnidex Administrator and OdxSQL also allow SQL statements to be directly issued to the database.
A variation of this architecture has the application establish two database connections - one to Omnidex and one directly to the database. The Omnidex connection is used when the application needs additional flexibility and performance. The relational database connection is used where Omnidex is not needed, or where the application wishes to use proprietary features of the relational database.
These architectures are commonly used for production applications with a light or medium query load. Once the query load becomes high, Omnidex is usually assigned a separate server from the application and the relational database. Once the database grows beyond 20 million rows, it is also common to use an Omnidex Grid to insure performance.
These architectures are also used when an Omnidex Snapshot is given to an individual for private use. Once Omnidex is installed on the analyst's computer, the Snapshot and its indexes can be distributed by simply copying a directory of files. Snapshots can reside on a database administrator's test server, an analyst's desktop or an executive's laptop.
More complex applications usually involve more servers, as shown in this Multiple Servers system.
See also: