Vous êtes ici

IBM Cognos BI Performance Tuning

IBM Cognos Performance Tuning

Performance issues and slow response times are often difficult to solve. The performance of a Business Intelligence and Performance Management platform is however a key success factor for improving user acceptance and buy-in for the delivered solutions.

It is crucial to reach and maintain a high level of performance for the IBM Cognos solutions. Each of the components must be configured and tuned to offer the best and most stable performance level possible with the resources available. But the number of components used, the multi-tier architecture and the dependencies with external components or factors make of the tuning exercise to be sometime more an art than a science. It requires a large set of skills including advanced knowledge of the IBM Cognos Business Intelligence and Performance Management components and optimization techniques.

Within the elementary methodology, element61 has defined a methodological approach to analyse and address performance issues in a Business Intelligence and Performance Management environment. Based on this methodology and on their past experiences of similar IBM Cognos tuning, our consultants will review the following components of your architecture.

  • Web-client and network
  • Web-tier : one or more web-servers
  • Application tier : one or more application-servers
  • Data tier : one or more data sources

They will determine where the main bottlenecks, if any, are, and will suggest and/or implement possible remedies to solve them and improve the level of performance of your IBM Cognos platform.

As an example, in the case of a Cognos Business Intelligence environment, the following detailed data-tier investigations will be done to check if the connections are correctly configured and/or correctly dimensioned :

  • Connection to the Content Store (JDBC)
  • Connection to Query databases
  • Connection to Cubes
  • Connection to Metric store (JDBC)

Experience shows that most performance problems are situated in the data tier layer, and basically often boil down to : "The generated queries run slow on the database”. Many different reasons can be the root cause for this :

  • The wrong queries are generated
    • Bad structure of the query objects in the Cognos reports
    • Bad structure of the framework model
    • The underlying database is not a data warehouse
    • The data model of the underlying data warehouse is not structured well
    • The data model of the underlying data warehouse should be extended with aggregation tables
  • The correct queries are generated, but they don't run fast
    • The wrong indexes have been created
    • The physical properties of the database are not optimised for a data warehouse environment
    • The database server is shared by too many applications, or by a mixture of OLTP and OLAP applications
    • The database server is dimensioned too small

Beside the tuning of a Cognos BI environment based on relational databases, element61 also has experience with the tuning of an environment that uses cubes as a main source, including:

  • Tuning of the cube model in order to shorten the cube build time
  • Tuning of the cube structure to improve the response time of queries that run on top of the cube
  • Tuning of the query objects inside complex reports that are built on top of cube, in order to make sure that performant queries are generated

Contact us to learn more about our IBM Cognos Performance Tuning Service Offering !