Schema on Read vs Schema on Write in Business Intelligence
When starting build out a new BI strategy. There are more options now than ever before. Gone are the days of just creating a massive star or snowflake schema to store THE data warehouse. In today’s business, you need to store data in structured and unstructured formats and deliver BI value to the users as quickly as possible. This is where the schema on read and schema on write comes into play.
Schema on Read
When using schema on read, you are storing you data in a non RDMS database structure. This allows you to quickly store any type of document or data required for analysis. The flexibility gained by having a limitless format to store any data is trading for speed of a structured RDMS table structure.
Schema on Write
A traditional data model requires the data to be parsed and loaded into specific tables and columns. This makes the load operations slower than schema on read, however, the ability to query pre-designed tables that can contain indexes makes the queries faster.