Transformation Metric with multiple transformation levels calculated using the lowest level after upgraded to Data Engine version 2019

Symptom

When a report uses a Transformation Metric with multiple transformation levels(e.g. [Last Year's], [Last Quarter's], [Last Month's]), Engine will pick the best fact tables with metric level. Since Strategy 10.11 and Data Engine Version 2019, Engine picks fact tables with lower levels, e.g., when Year level fact table is available and can be used, the Engine picks a table at the Day level. Both SQL and Data could be affected by this issue. 

Reproduce

• Use Strategy Tutorial Project
• Create a Transformation Metric M1 with multiple transformation levels

• Create a Report and put Year and M1 on the template
• View SQL:
  
• In Data Engine 10.4 (join at year level table: yr_category_sls)

• In Data Engine 2019 (join at month level table, mnth_category_sls)

Cause

When there is more than one Transformation level, the Data Engine incorrectly picked an attribute that is lower than the optimal choice and thus could yield less optimal data and worse performance.
This is a known issue starting from Strategy 10.11 with Data Engine Version 2019 or above.

Solution

To fix the data issue, upgrade to Strategy 2021 Update 5 and set the Data Engine Version to be 2021. To set Data Engine Version, refer to How to set Date Engine Version.
After the fix, the behavior will be reverted back to the same as Data Engine 10.4
Strategy Internal Notes:
Strategy Internal Reference number for this technical note is DE205728.