1:1 mentoring with Big Tech AI engineers
System Design

Where It Runs

Which fields the platform really owns, which runtime shape fits an agent's run length, and why the front door's timeouts decide what a 75-second run can do.

Last updated

Production10 min readFirst readService and Container

After this section you can

  • Name which object owns each field of the deploy contract
  • Pick a runtime shape from the run's length and the timers in front of the service
  • Roll out a change without losing capacity: a startup probe, surge and unavailable limits, a PodDisruptionBudget
28

Where It Runs

Which object owns each field of the deploy contract, which runtime shape fits a run's length, and what the front door's timers allow.

Key idea

Every platform asks the same few questions about a service, and a run's length decides its shape. A 75-second run cannot sit behind a 29-second gateway as one synchronous call: it streams, queues or runs durably.

Run length picks the shape, and the front door decides which shapes are possible
RUN LENGTH, LOG SCALE FAQ action investigation 1 s 10 s 75 s 4 min hours 29 s, the gateway limit A · Request and response a caller waits B · Queue and workers a queue, minutes D · Durable workflow outlives a deploy, waits for a human C · Scheduled job, started by a clock

Blue: a caller waits. Teal: queue and workers. Purple: state. Amber: model work. Red: the gateway limit. Pans sideways on a phone.

Related

More in System Design

Get full access to all 74+ sections with code examples, diagrams, and interactive animations.

Unlock Premium