Cognos Dynamic Cubes provide increased scalability and a richer set of dimensional modeling options

screenshot_1
First, here is a quick list to describe Cognos Dynamic Cubes:

  • Dynamic cubes (DC) are in-memory OLAP (on-line analytical processing) cubes that load data directly from relational data warehouses.
  • They introduce a performance layer in the Cognos dynamic query stack to allow low-latency, high-performance OLAP-style analytics over large relational data warehouses.
  • This is built into the Dynamic Query Mode server – no separate installation or licensing needed.
  • Each cube represents a dimensional view of a star or snowflake schema, and each is based upon one fact table, or several aggregate tables.
  • Dynamic Cubes leverage the power and scale of a relational database, including aggregate tables, to provide OLAP-style analytics relational data.
  • Once a DC is created and published as a package, it can be accessed and used for reporting just like any other OLAP package (QS, AS, CWA, RS).
  • DC requires a cube designer, not a framework manager.

Two of the most interesting points regarding this new cube feature in Cognos are:

  • Aggregate fact tables can be integrated into the base cube model.
  • The data comes directly from a real data warehouse, and is not just a snapshot.

Furthermore, it is easy to work around the limitation of modeling each cube on one fact table in a star/snowflake schema by creating a virtual cube, which is the merger of two real cubes or one virtual and one real cube. In other words, with DC you can work with any data warehouse and also create an aggregate aware model to provide OLAP-style analytics against dimensionally modeled relational data (DMR) on the newly introduced 64-bit Dynamic Query Mode (DQM) service in addition to the traditional 32-bit Compatible Query mode, which has been a default for all Cognos versions up to 10.x

Of course, we need to know at the same time that, in order to enable DQM on FM models or packages, we should use valid data sources which are supported for DQM by Cognos. First you should have the correct JDBC drivers installed, plus the following data sources should be used:

  • OLAP perspective: IBM Cognos TM1, Microsoft SSAS, Oracle Essbase, SAP NetWeaver, SAP BW;
  • Relational concept perspective: DB2, SQL Server, Netezza, Oracle, Teradata, Informix, IMS

So, now in Cognos BI you can have three types of OLAP options for your business analysis. Below is a comparison of the pros and cons of each.

OLAP
options                             Pros                                                                          Cons

Dynamic Cubes High data volumes Star or snowflake schema requires
Low latency/fast performance/64-bit
Optimized aggregates/aggregate aware
FM DMR Leverage existing Framework Model Low/medium data volumes
Caching for performance (DQM)
Performance slower using transactional systems
Power Cubes Data moved into physical cube structure Low/medium data volumes
Latency linked to cube build / refresh times
32-bit (2GB size limit)

DC provides increased scalability while enabling users to share data caches and create a DC data source pre-loaded with dimensions. This richer set of dimensional modeling options:

  • Eliminates size limits
  • Allows use of a 64 bit DQM service to process requests faster
  • Allows caching of popular parts (IE: slices and dices) of the DC, making it unnecessary for Cognos to go back and aggregate already existing counts, etc.