Part 1 of 8 · 1 min read
Why this page comes after indexes
In this module you will
- Describe how a B-tree finds a key and how it grows by splitting pages, and compute how many levels a given amount of data needs.
- Describe how an LSM-tree handles a write with a memtable, a log and sorted files, and how it finds a key again.
- Walk through a compaction on small files, including a deleted key.
- Define write amplification and compare the two designs with numbers.
- Say why a Bloom filter helps an LSM-tree.
- Choose between the two for a workload and name the product families that use each.
The previous page treated an index as a tool. This page opens it. In an interview, a deep question such as "why is Cassandra good at writes?" or "why does Postgres use a log before the table?" is answered here. The tier is Common: not every role needs it, but it explains many choices made in later levels. The numbers on this page are made up to show the idea, unless the text names a book.