The architecture of IBM Cognos Dynamic Cubes is something that many find it hard to conceptualize. Cognos Dynamic Cubes are really just an extension of Dynamic Query Mode (DQM) which in turn is part of the three tiered Cognos BI architecture familiar to authors, modellers and administrators alike. So why is it that as we delve into the Cognos BI stack tracing these extended features it becomes difficult to stay on track?
I find that the trick is to separate the concepts of Cognos Dynamic Cubes as a cubing engine and dynamic cubes as a data asset (or reporting data source). For those familiar with TM1, a similar distinction can be found between the TM1 server and the memory-resident cubes it hosts. In the case of Cognos Dynamic Cubes, the cubing engine is presented as the Query Service and the running cube instances reside in the associated Query Service JVM.
A further challenge for those familiar with the Cognos BI architecture is that unlike other reporting data sources accessed by Cognos BI, a dynamic cube can have multiple running instances which reside entirely inside the Cognos BI software stack. So, what is created when you publish a cube model and what are you connecting to when you open a dynamic cube package from one of the Cognos BI studios?
To answer this question I like to start with the physical components in a dynamic cube. By physical components I mean the actual bits and pieces that reside in the Content Store database. Extending my comparison with TM1, here I would be talking about the data files where the cube dimensions, rules, leaf level data, and so on required to load a TM1 cube into memory are stored. Coming back to Cognos Dynamic Cubes, the corresponding component is the cube model, but wait … there is more.
First, there is the published cube model. This object contains the cube specification (foe example, relational metadata, dimension structures, measure mappings, in-database aggregate specifications, virtual cube specifications, security views, and so on). The published cube model does not contain any cube data and is not accessed directly by the reporting studios.
Next is the dynamic cube data source connection. This object contains the connection information required by packages to access the dynamic cube. It also has a property, known as the access account, which stores the credentials used to access the cube's underlying database. The dynamic cube data source connection is created automatically on publish, it cannot be constructed manually.
Finally there are the cube configurations. Cube configurations specify which dispatchers can host a running instance of the cube and include properties such as size limits for the cube caches.
So putting these pieces together:
- A dynamic cube, defined in IBM Cognos Cube Designer, is published to the content store in the form of a published model and a data source connection (with an access account property).
- Dynamic cube configurations are created when the cube is configured against each dispatcher.
- The published model, access account and cube configurations are used by Cognos Dynamic Cubes to build the various running instances of the cube in the appropriate Query Service JVM's and populate the caches of each cube instance from the underlying data source.
- The single dynamic cube data source connection is referenced by Cognos BI packages thus making the running cube instances available to the Cognos BI studios.
To find out more about where the data in a cube comes from, how it gets there and what is involved in managing your cubes take a look at new 2nd edition of the IBM Redbooks publication IBM Cognos Dynamic Cubes, SG24-8064-01. For a high level overview of IBM Cognos Dynamic Cubes see the IBM Redbooks Solution Guide Big Data Analytics with IBM Cognos Dynamic Cubes, REDP-5265
As both a consultant and educator, I am often asked for help with the transition from the training room to real business processes. I have found the IBM Cognos Dynamic Cubes book an invaluable resource in bridging this gap and recently had the privilege of working with a team of Cognos experts from IBM to co-author this latest edition.
To learn more about IBM Cognos Dynamic Cubes performance, take a look at the blog post by Avery HagleitnerIBM Cognos Dynamic Cubes: Improving performance of reports and dashboards
MaryAlice Campbell is a Senior Consultant and Business Analytics Technical Practice Leader at ISW, Australia. She has over 20 years of experience as a business analytics specialist. MaryAlice is an IBM Cognos BI veteran having gained experience with the early, pre-web versions of IBM Cognos PowerPlay and IBM Cognos Impromptu®; she contributed to beta and training programs, and worked with all subsequent releases. MaryAlice is also an IBM Certified Solution Developer, internationally recognized educator, and a Master Instructor of the IBM Analytics curriculum. MaryAlice is one of the authors of the IBM Redbooks publication IBM Cognos Dynamic Cubes, SG24-8064-01 and the IBM Redbooks Solution Guide Big Data Analytics with IBM Cognos Dynamic Cubes, REDP-5265.
Likes before 03/04/2016 - 1
Views before 03/04/2016 - 3722