Posted on Jul 16, 2019

Resource efficiency for Complex planning models

Cognos TM1 / Planning Analytics, is a powerful in-memory, online analytical processing (OLAP) database technology used to model and process large sets of data.  The database uses a calculation and consolidation engine that together process complex calculations and feeds the results to multiple cubes within the database as well as other databases as required.

While the design of the database allows for good performance on cube calculation results ‘out-of-the-box’, the performance of a Planning Analytics system can be significantly enhanced through code and cube design as well as the efficient management of feeders; essentially minimising over-feeding and avoiding under-feeding.  Bold Shore developers have managed to achieve this by using one or more of the following techniques:

  • Adopting ‘Global’ cube rules for the management of planning versions – this technique allows one to preserve plan data without making any further ‘calculation’ demands on server resources
  • The use of multi-threaded queries in both the rule engine as well as in the Turbo Integrator (ETL) layer – this maximises utilisation of available resources and thereby enhances overall system performance
  • The use of Configuration cubes, and feeding from these cubes – this allows for system configuration flexibility without the need to continuously alter code when changes are required, while still achieving efficient feeders

The result of efficient rules and feeder statements is a database that provides enhanced performance and maintains the flexibility that allows users to configure their system, but which requires less memory and CPU cycles to run.  By utilising these techniques, Bold Shore has managed to create complex planning analytics models that have a relatively small hardware resource requirement when compared to a typical planning system.

A further method of creating resource efficiency in a planning system solution is to adopt a distributed architecture – both in terms of database and web layer distribution, as well as the creation of multiple database instances for different operations within a group of companies.  The database and web layer distribution allow for the separation of the database calculation load and web server processes across multiple servers, while different instance distribution ensures that individual model sizes remain manageable and capable of calculating results efficiently.

In order to mitigate any perceived risk and/or workload to maintain multiple database instances, Bold Shore utilises Task Automation software to monitor log files in different locations that generate notifications if user-related or system error messages are detected, allowing for a proactive approach to issue resolution.

From the planning of the infrastructure on which the Planning Analytics software runs, through the implementation of Cube Rules, Feeders and TI processes, to the use of specific 3rd party applications, Bold Shore provides an efficient, powerful and flexible planning solution that allows end users to Plan and Analyse Budget and Forecast data with ease.

By Murray van Rooyen

Tags: , ,