Friday, September 24, 2010

data warehouse dimensional models?

star schema model
snowflake schema model

for performance reasons as well as design simplicity and consistency, it is better to stick to a star schema rather than a snowflake, at least as much as we can.

A star schema is simpler and is more consistent than a snowflake schema because it has only one level in all dimensions. Because a star schema is simpler, it is easier for the ETL processes to load the data into it.

A snowflake schema is when you normalize a dimension into several smaller tables with a hierarchy structure. It is used when you want to benefit from less data redundancy.The benefit of a snowflake schema is that some analytics applications work better with a snowflake schema than a star schema.

The other benefit of a snowflake schema is that less disk space is required. A snowflake schema can decrease query performance, but it can also increase query performance. It decreases the fact table query performance because we have to join more tables to the fact table. It increases the query performance when we want to get the distinct value of a particular dimension attribute. If this attribute is normalized
into its own subdimension table, the select distinct query runs more quickly.

No comments:

Post a Comment