Using a Distributed, Columnar Database to Improve Analytics - Database Trends and Applications
The increased requirements of modern analytical workloads, querying billions of rows on demand, are a challenge for relational databases because they're optimized for transactional workloads. While transactional workload queries tend to be row-oriented (e.g., return every column in a single row), analytical workload queries tend to be column-oriented (e.g., return the aggregate of a single column in every row). By storing columns of data rather than rows of data, columnar databases optimize for analytical workloads without sacrificing the relational model and SQL.
Comments