Data modelling

Normalization

• normalized

  • reduce data(field) duplication

  • data storage is more efficient for write performance

  • simple queries run faster

  • data is scattered across multiple tables

• denormalized

  • duplicate data across tables to improve read performance over write performance

  • data exists in 1 table, as opposed to multiple tables

• check https://en.wikipedia.org/wiki/Database_normalization#Normal_forms

Transactional Schema (OLTP)

• normalized schema

• used for OLTP - transactional workloads

Star Schema (OLAP)

• easiest schema design for data warehouse/mart

• data is organized into a central fact table that contains the measures of interest, surrounded by dimension tables that describe the attributes of the measures.

• used for OLAP - analytical workloads

• fact table: usually numeric values that can be aggregated

• dimentional table: groups of hierarchies and descriptors that define the facts (attributes)

• data is joined through FKs

• data can be denormalized, to improve reporting reads

• fact table examples: sales transactions, weather measurements

• dimension table examples: client info, seller info, product info

Snowflake Schema (OLAP)

• snowflake schema is a generalization of a star schema

• similar to star schema, but dimensions are normalized into multiple related tables (linked through FKs)
• some database developers compromise by creating an underlying snowflake schema with views built on top of it that perform many of the necessary joins to simulate a star schema.
• Some OLAP multidimensional database modeling tools are optimized for snowflake schemas.[3]
• Normalizing attributes results in storage savings, the tradeoff being additional complexity in source query joins.