InetSoft Documentation: Distributing Materialized Report Views

Users of InetSoft's reporting software can greatly improve the performance of reports by allowing users to distribute materialized report views that vastly improve report run-time. View the information below to learn more about the Style Intelligence solution.

A materialized view is a construct in which data is pre-aggregated and cached, enabling Viewsheet results to be more quickly computed at runtime. When a materialized view exists, the report engine attempts to query the cached view rather than the original database tables, which may significantly reduce the run time for certain queries.

Only Viewsheets that contain aggregated data can be considered candidates for materialized views. For best performance, aggregation should generally be performed in the worksheet rather than in the Viewsheet. To improve performance, you can distribute a materialized view across multiple machines. Materialized views are available only for Viewsheets.

Materialized views can be managed from the 'Repository' > 'Materialized View' node under the Server tab. The following sections provide more information about management.

You can manage existing materialized views from the 'Materialized View' page under the 'Repository' node of the Server tab. The 'Delete' button deletes the selected materialized view, which may be useful if changes have been made to the underlying database schemas.

You can improve Viewsheet performance substantially by distributing materialized views across multiple machines or nodes. In this case, the materialized dataset that Enterprise Manager generates for the Viewsheet is saved in its entirety on each individual node. This allows the Viewsheet server to pull small blocks of data from each of the nodes, minimizing the input-output demands (hard disk access) on a particular machine.

The reduced input-output load allows the Viewsheet to be generated and updated much more quickly. For this reason also, best performance is achieved by using a large number of average machines, rather than a small number of very powerful machines.