Why Databases Use B+ Trees Instead of BSTs
Why Databases Use B+ Trees Instead of BSTs
In RAM, a balanced Binary Search Tree (AVL, Red-Black) gives O(log₂ N) search time. But on disk, O(log₂ N) means 20 to 30 random disk I/O seeks per query!
⚡ The Fan-Out Advantage
A B+ Tree node is sized to match the storage block (typically 4KB to 16KB). With a fan-out of 512:
- Level 1: 1 page (Root)
- Level 2: 512 pages
- Level 3: 262,144 pages (~134 million keys)
Search requires only 3 disk reads instead of 27!
Read the complete storage lesson on Binary Dose.
原文：https://www.a8q9.cn/en/article/why-databases-use-b-trees-instead-of-bsts-be3516.html