Skip to main content
José David Baena
Tools and working references

Background jobs / Capacity calculator

Backlog Drain-Time & Worker Capacity

Can this queue clear before the deadline, or does incoming work consume all available capacity?

Interactive calculations run in your browser; the initial example is pre-rendered. There are no accounts, uploads, or live queue connections. Inputs stay in page memory; the site does not persist them, put them in URLs, or send them to analytics. A worksheet download includes only what you explicitly export.

Set the recovery model

Set the recovery model

Jobs waiting at the start of the model.

Assumed constant arrival rate during recovery.

How long a job occupies one slot, not a latency percentile.

The model assumes identical workers; zero means no active workers.

Simultaneous jobs per worker, not necessarily CPU cores.

Blank means no modeled cap; zero models a stopped dependency.

The target is a cleared backlog at this time.

Capacity estimate

The backlog drains, but misses the deadline

Effective capacity

16 jobs/s

Limited by worker slots

Net drain rate

6 jobs/s

Capacity minus new arrivals

Drain time

33.33 min

Steady-state estimate, not a promise

Workers for deadline

9

Total workers, not additional workers

Worker-only capacity
16 jobs/s
Capacity required by the deadline
16.67 jobs/s
Backlog at the deadline
1,200 modeled jobs
Arrival load / capacity
62.5%

The suggested worker count meets the arithmetic target, but adds no safety margin for bursts, variance, or worker startup. If capacity exactly equals arrivals, there is no room to absorb a new backlog.

Inspect the arithmetic

worker capacity = workers * slots / mean service time

effective capacity = min(worker capacity, downstream cap)

drain time = backlog / (effective capacity - arrivals)

required capacity = arrivals + backlog / deadline

Assumptions and limits

  • Constant arrival rate and mean service time; each job occupies one slot until it completes.
  • The downstream limit is effective jobs per second, not raw API requests per second.
  • Workers scale linearly until the stated downstream limit; startup, contention, retries, and failures are not modeled.
  • This fluid estimate does not predict latency percentiles or guarantee a deadline under bursty traffic.