Scheduling, Capacity, Cost, and Fairness
Govern scarce model, compute, environment, tool, and human review capacity.
A focused view of boundaries, contracts, state, authority, failure paths, and tradeoffs drawn from this chapter.
Reconstruct and defend this chapter’s architecture.
Reconstruct the architecture, name each boundary, and defend the tradeoffs.
Open the source exercise
Schedule an incident, routine feature, security update, and large evaluation across constrained model quota, two sandbox pools, and one reviewer. Add a provider outage and retry storm. Explain every queue and preemption decision.
4. Tradeoffs and alternatives
Sophisticated schedulers improve utilization and are hard to explain. Begin with explicit priority classes, quotas, concurrency, aging, and reserved capacity. Predictive duration helps packing but can disadvantage novel work. Cost limits prevent runaway use and may block valuable investigation; provide scoped escalation with owner and expiry.
5. Current Mission Control Implementation
The current architecture includes queues, leases, worker capabilities, budgets, model routing, provider rate limits, concurrency, and health metrics. These support bounded execution.
The curriculum does not yet specify a complete admission and scheduling policy, fairness model, preemption protocol, capacity forecast, reviewer-capacity constraint, or end-to-end cost attribution. Existing economic metrics require this operational layer to become actionable.
Review this chapter.
Challenge a claim, boundary, missing failure mode, unclear term, or unsupported evidence statement.
- Claim
- Boundary
- Failure
- Evidence