Time series optimizations in network BI best practices

When using Network BI (through either Packages or Networks), different facts can be coming from different parent spaces or be generated in the child space.

For example:

  • Model Space A contains a "Products" fact that contains the count of CategoryID.
  • Model Space B contains a "Shippers" fact that contains the count of ShipperID.
  • Child Space C imports "Products" from Space A, "Shippers" from Space B, and contains a local fact "Orders" that contains measures on Sales and Quantity.

In Network BI scenarios, we recommend for all spaces in the architecture to contain the exact same Time Series optimizations. This means that if you need to be able to use the TTM (Trailing Twelve Months) for the CategoryID measure or QAGO for the ShipperID measure, you must ensure both TTM and QAGO are set up as enabled for Space A, B, and C. We recommend you add the exact same time period levels across all your spaces.

Note: This ensures your queries will function in your reporting spaces and also for performance optimization since the child network space will be initializing the repository of parent spaces. Optimizing the parent space's virtual measure tables is beneficial for the performance of the child space.
Caution: 
It is possible to only optimize Time Series in children spaces. This results in easier maintenance and provides performance gains. However, the Time Series queried through the child space must always be present in the parent space to guarantee accurate queries to be generated.