Backpressure is the depth

Before you start

• Queues

  Work moves off the request path into a queue.

  Open lesson

Why it matters

A normal author can have hundreds of followers, and each home timeline needs the new post within a few seconds. If the post request waits for every timeline update, its latency grows with follower count and bursts make authors wait for overloaded workers. The post API must return once the post is durable, regardless of follower count, while timeline lag stays within the freshness promise.

A normal user posts text. Followers’ home timelines should include it within a few seconds. The post API should return as soon as the post is durable.

Input

arrivals = [800, 1200, 1500, 900, 600, 400], capacity = 1000

Output

[0, 200, 700, 600, 200, 0]

For a normal post, commit once and enqueue its post and author ids, then let workers write the followers’ timelines; measure the 800, 1,200, and 1,500-message burst against 1,000/s capacity so post latency never follows fanout depth.

After the post row commits, publish its id and author id. A worker expands that message into one timeline write per follower. A follower who opens home before the worker reaches them sees the previous page; the new post appears after the agreed few-second gap.

Track queue depth each second as prior depth plus arrivals minus the 1,000-message worker capacity. In the example, bursts raise depth to 700 and spare capacity drains it to zero. Rising lag triggers more workers or an alert; sustained overload needs admission control because a queue cannot erase excess work.