Scaling Data
The moves you make when one database stops being enough — and the consistency you give up for each one.
6 topics
Replication & Read Replicas
Replicas multiply read capacity and introduce lag, which shows up as a user not seeing their own write.
core20 minPartitioning & Sharding
Splitting one big table across partitions or machines, and why the key you choose is the decision you can't undo.
core25 minDenormalisation & Materialised Views
Precomputing an expensive read, and the three ways to keep the precomputed copy honest.
core20 minHot Rows & Write Contention
When every request updates the same row, throughput collapses to one transaction at a time — and the fixes are all forms of spreading the write out.
deep20 minN+1 & Query Patterns
The single most common performance bug in application code, why ORMs and GraphQL make it easy, and the batching fix.
core20 minOutbox & Dual-Write Consistency
You cannot atomically write to your database and send a message, so write the message to the database and publish it afterwards.
deep20 min