(Pt. 3) Postgres vs MySQL: HOT Updates
Part 3 of a five-part series: Clustered index vs heap MVCC + vacuum vs undo log HOT updates (this post) Page splits, TOAST, and UUIDs Connections: processes vs threads I...
Seiji's Eng Blog
(Pt. 2) Postgres vs MySQL: MVCC + Vacuum vs Undo Log
Part 2 of a five-part series: Clustered index vs heap MVCC + vacuum vs undo log (this post) HOT updates Page splits, TOAST, and UUIDs Connections: processes vs threads I...
(Pt. 1) Postgres vs MySQL: Clustered Index vs Heap
A five-part series on how PostgreSQL and MySQL (InnoDB) differ under the hood: Clustered index vs heap (this post) MVCC + vacuum vs undo log HOT updates Page splits, TOAST, and U...
Designing DynamoDB Keys: PK, SK, LSI, and GSI
A Scan with FilterExpression on Status = PENDING returned ten orders. ScannedCount came back 400,000, and at roughly a kilobyte an item that’s 400 MB read to hand back ten items, or about 50,000 re...
How PostgreSQL Builds Indexes Without Blocking Writes
CREATE INDEX takes a SHARE lock for the entire build, so every write to the table waits. On a large table that’s minutes or hours of rejected writes, and adding CONCURRENTLY makes it go away. The ...