An interactive study companion for Martin Kleppmann's classic on reliable, scalable, and maintainable systems.
Leader-based replication funnels all writes through one node and copies them to followers.
Replication keeps copies of data on multiple machines for fault tolerance and read scaling. The simplest scheme designates one leader that accepts writes and broadcasts changes to follower replicas.
PostgreSQL streaming replication sends the WAL from primary to standbys. Amazon RDS Multi-AZ runs synchronous replication within an AZ pair. MongoDB replica sets elect a primary; secondaries tail the oplog. Read replicas on AWS Aurora offload analytics queries from the writer.
Message history replicates from a primary shard to followers so reads scale globally. All writes funnel through the leader; followers tail the WAL to stay consistent without accepting direct writes.
// WhatsApp-style leader → follower replication via write-ahead log
async function replicateWrite(entry: WalEntry) {
await leader.appendWal(entry);
for (const follower of followers) {
follower.streamWal((e) => follower.apply(e)); // async tail of leader log
}
return leader.confirmWrite(entry.lsn);
}