Part 1 of 6 · 1 min read
Why this page comes first
In this module you will
- Explain the difference between row-oriented and column-oriented physical storage.
- Explain why a typical analytical query only touches a handful of a fact table's many columns.
- Describe how bitmap encoding and run-length encoding compress a column's values.
- Explain how a chosen sort order affects both query speed and compression.
- Explain why column-oriented storage makes writes harder, and how an LSM-tree-style approach fixes that.
Earlier pages in this level covered the warehouse's schema and the engines that process huge batches and streams. This page covers the physical shape the warehouse's own fact table is actually stored in. That shape is the detail that makes a scan over billions of rows practical at all.
Remember this
Row-oriented storage keeps one row's columns together; column-oriented storage keeps one column's rows together, and a typical analytical query only needs a few columns.