Principle

Push back on intake before the backlog becomes the outage

A growing queue is a message from capacity, not a storage success story.

When it fits

  • Work arrives faster than it can be completed and producers keep submitting at the same rate.

When to avoid it

  • Backpressure needs clear user behavior; invisible throttling can look like random failure.

Why it matters

When consumers approach their sustainable limit, slow or reject new intake rather than allowing unbounded backlog growth. Propagate the pressure to the producer through quotas, concurrency limits, retry-after signals or explicit scheduling. The producer must see that capacity is constrained.

An example

Limit new agent jobs when human review capacity is saturated instead of generating thousands of outputs that will expire unread.

Check your result

Backlog pressure changes intake behavior before the system exhausts memory, deadlines or reviewer capacity.

Keep this limit in mind

  • Backpressure needs clear user behavior; invisible throttling can look like random failure.

Connected ideas

Useful with
Give shared capacity a noisy-neighbor rule

Evidence and sources

Supports

AWS's 2026 fairness guidance uses throttling, quotas and backpressure to prevent one workload from starving others in a shared system.

Fairness policy must reflect actual priorities rather than equalizing workloads that have different importance.

Fairness in multi-tenant systems · Admission control and fairness

Supports

Google SRE recommends small queues or early rejection under steady overload because long queues add latency, memory use and work that may already have missed its deadline.

Bursty asynchronous workloads can legitimately benefit from larger queues when backlog remains useful.

Addressing Cascading Failures · Queue management and latency/deadlines

All sources (2)