After data is loaded into Snowflake, it automatically reorganizes the data into a compressed, optimized, columnar format (micro-partitions). The optimized data is then stored in the cloud storage. Snowflake manages all aspects of storing these data, including file structure, size, statistics, compression, metadata, etc. Snowflake data objects aren't visible to customers or users. Users can only access data by performing SQL queries on Snowflake. Snowflake uses a columnar format to optimize and store data within the storage layer. With it, data is stored in columns instead of rows, allowing for an analytical querying method and improving database performance. With Columnar databases, business intelligence will be easier and more accurate. Compared to row-level operations, column-level operations are faster and use fewer resources than row-level operations.
Above is an example of a table with 24 rows divided into four micro-partitions, arranged and sorted by column. As the data is divided into micro-partitions, Snowflake can first remove those micro-partitions not relevant to the query, followed by pruning the remaining micro-partitions by column. The result is fewer records traversed, resulting in significantly faster response times.
