Data storage options for uploaded files and data sets

Cognos Analytics can store the data of each uploaded file or data set in one of two formats:

  • tabular data, stored in a Db2 database
  • Parquet files, stored in the Content Manager database

Starting with IBM® Cognos® Analytics 12.0.2, the data can persist in a supported Db2 instance that is configured for analytic workloads. For more information, see Storing data from uploaded files and data sets in Db2.

Prior to Cognos Analytics 12.0.2, data in uploaded files and data sets was always stored as Parquet files in a Content Manager database or an external Content Manager store. This storage option is still available, and is the default option when there is no Data sets store connection that can be used. For more information, see Data types used to store data from uploaded files and data sets in Parquet files.

The Data location property on uploaded files and data sets shows the location of the data. If Db2 is configured as the data storage location, this property shows the name of the data server connection to the configured Db2 instance or schema. Otherwise, the property shows the default location, which is Content Manager.

Note:

If you are a Cognos Analytics on Cloud customer who wants to use the storage option for uploaded files and data sets, you must purchase a part to have this feature enabled.

The database is managed and configured by the IBM SRE team.

If you do not purchase a part, you can continue to store your uploaded files and data sets in Parquet files and use the Compute service.

Differences between Parquet and Db2 storage

Parquet data stored in Content Manager is limited to 2GB per Parquet file.

Data stored in Db2 must comply with the following Db2 limitations:

  • Character columns cannot have a precision greater than 8192.

    Use the CAST expression to reduce the precision of the source value.

  • Decimal columns cannot have a precision greater than 31.

    Use the CAST expression to reduce precision, if required. Or use a CAST expression to DOUBLE if approximate numeric value is acceptable.

  • Total row size of all columns cannot exceed the Db2 row limit.

    Use CAST expressions to reduce the precision of columns or drop columns, as applicable.

Data stored in Db2 is limited by the available space in the tablespace where tables are created.

For more information, see the Db2 documentation:SQL and XML limits and Data types.