An interactive study companion for Martin Kleppmann's classic on reliable, scalable, and maintainable systems.
Asynchronous followers may lag behind the leader, creating stale reads and subtle consistency bugs.
When followers lag seconds behind the leader, users experience anomalies: posting a comment and not seeing it, or seeing newer data followed by older data. These are not failures — they are consequences of eventual consistency.
Read from leader after writes, use sticky sessions, or wait for follower catch-up before serving reads.
After posting a tweet, Twitter routes your next read to the leader or a caught-up replica. E-commerce apps using PostgreSQL read replicas often force read-your-own-writes by querying the primary after checkout. HAProxy or AWS ALB sticky sessions keep a user on one replica for monotonic reads.
After you post a story, the next feed load must include it — even if replicas lag milliseconds behind. Meta routes post-write reads to the primary or enforces a short read-from-leader window per session.
// Facebook / Instagram: route reads to primary after posting
async function createPost(userId: string, text: string) {
const post = await db.primary.posts.create({ data: { userId, text } });
// Sticky session: next reads hit primary or caught-up replica
session.set("readFromPrimaryUntil", Date.now() + 5_000);
return post;
}
async function getFeed(userId: string) {
const client = session.get("readFromPrimaryUntil") > Date.now()
? db.primary
: db.replica;
return client.feed.findMany({ where: { userId } });
}