Cognos Analytics with Watson

CA on Cloud: Transformer cube build performance considerations

By Luke Peterson posted Wed February 20, 2019 11:44 PM

Transformer cubes are supported in Cognos Analytics on Cloud (CAoC). Cubes are built on premise and then deployed to the cloud for consumption by end users. In addition, Cognos Analytics on Cloud currently supports only DQM packages. (reference: )

If your Transformer cube builds are based on Framework Manager models or reports that have been published to Cognos Analytics on Cloud, your packages will have been migrated to DQM as part of the move to cloud.
Consider this Transformer product functionality:
• When Transformer model source is a package or report using DQM, data access occurs from the Query Service on the CA server.
• When Transformer model source is a package or report using CQM, data access occurs from the Transformer build machine itself.

If your datasources are on-premise, cube build times may be reduced by leveraging the data access layer built into Transformer. By basing Transformer cube builds on a CQM package and datasource, your on-premise Transformer application can retrieve data from your on-premise datasource without moving the data through the CAoC reporting server. This will reduce the number of hops and leave the CAoC datasource connection fully available for reporting purposes.

To utilise this technique, you will need to have database drivers available on the Transformer cube build machine (since the connection is CQM you will need the 32bit full client and configuration). Along with the drivers you will need to publish your package in CQM and create a CQM datasource. Then modify your Transformer model to use the CQM package and data access will now occur from the Transformer machine. (If your Transformer source is a report, change the report properties to point to the newly published CQM package.)

Of course, the CQM package that you have published cannot be used for reporting purposes, nor will the datasource test successfully from CAoC. To avoid confusion for consumers, consider hiding the package or placing it in a secured location where end users will not see it. (