Database

Replication

Statement-based replication - side effect like NOW()

Write-ahead log (WAL) shipping
- Undo log
- Redo log

Logical log replication

- MySQL’s binlog

Read your own writes

- read-after-write consistency

- When reading something that the user may have modified, read it from the leader, otherwise read it from a follower.

Monotonic reads

- time will not go backwards

- each user always reads from the same replica

Multi-datacenter operation

Multi-leader replication

- each datacenter’s leader replicates its changes to the leaders in other datacenters.

- dangerous

Handling write conflicts

- last-write-wins

- Record the conflict in an explicit data structure that preserves all information

- Conflict-free replicated data types (CRDTs)

Avoid Infinite replication loops

- each write is tagged with the identifiers of all the replicas it has passed through

Read repair

Anti-entropy process

sloppy quorums and hinted handoff

Traditional SQL
Alter table (add field, rename etc) is implemented by locking the original table,  creating a temp table, copying data, making change, then renaming the table.

Internals
How to implement auto increment _id in MongoDB
Use Counters Collection
Optimistic Loop

MySQL
- Lock

How to Use PreparedStatement and Why
PreparedStatement ps = conn.prepareStatement("select a,b from t where c = ?");
ps.setInt(1, I);
ResultSet rs = ps.executeQuery();