Part 1 of 5 · 1 min read
Why this page comes first
In this module you will
- Explain why a B+-tree's in-place updates become expensive under very heavy write load.
- Describe how an LSM-tree buffers writes in memory before flushing them to disk.
- Explain what an SSTable is, and why it is never modified once written.
- Explain what compaction does, and why it is necessary.
The previous two pages covered B+-trees and hash indexes. Both update a page in place when a row changes, as described back in Level 4's write-path page. This page covers a different family of structure, built specifically for workloads dominated by writes rather than reads.