Verdict
- The B+Tree is ~50 years old, still widely used, and essentially always the right choice for a database system
1:20:27. - The recommended build: classic B+Tree plus the B-link tree's sibling pointers plus the optimizations from topic 12 — that combination handles anything
1:20:27. - Building a new system from scratch? This is your first choice
1:20:27. - On skip lists: "I don't care what you read on Hacker News about skip lists, they're terrible"
1:20:27. - On tries: interesting, but you can put tries inside your B+Trees
1:20:54. - Opening claim, restated: the best data structure in computer science — "forget red-black trees, forget splay trees, nobody uses that"
0:27.
Next lecture
Other index data structures 1:20:54:
- Skip lists (and why they're a bad idea)
- Bloom filters
- Tries
- Inverted indexes
- Vector indexes (briefly, at the end)
Also deferred to the following lecture: latching and concurrent access for B+Trees 6:27 47:24.