Palo Community Forum

Hi,
Palo is not restricted to a certain source scheme, therefore there is no special manual for your approach. A Snowflake scheme is a good starting point. Some points you may have to consider are:
- transfer your "flakey" dimension tables back. So if you have normalized separate tables for city/region/country you may create a separate table that collects the information. From this table you could create
your geographical dimension including a hierarchy. There are a lot of examples in the Palo sample dbs and manuals. You may create a view layer for this on top of your DWH. Consider your Snowflake schema as a universal data provider that serves many frontends/data marts. With Palo you are probably connecting to a part of the DWH.
- distinguish dimension elements from (OLAP-)attributes
- regarding time dimension(s) you need to decide if you need one ( e.g. Year/month/Date) or two or more. There are some topics in this forum discussing this question.
- create a separate dimension containing the measures of your fact table. You can create hierarchies within this dimension as well.
- use readable element names instead of technical id's from your source system or concatenated names (id + name)

Take a look at the examples that come with Palo and you wil get ideas on how your final model should look like with respect to your requirements. HTH