How many mobile proxy ports do you need?
Size a mobile proxy workload using measured request time, concurrency, session needs and rotation headroom instead of a universal accounts-per-port rule.
Measure one port against your authorized workload, then estimate the port count from sustainable throughput and session requirements. Validate the result with a small increase. Unlimited bandwidth does not establish unlimited concurrency or guaranteed availability.
Describe the workload before counting ports
Start with the work that must finish in a defined interval: requests, average response size, browser sessions and their duration. Note whether requests can run independently or must stay in one session. A fleet that handles short API calls may behave differently from the same fleet running browsers and downloading images.
Keep target-side permission and rate limits in the model. More ports do not make an unauthorized workload acceptable or guarantee that a destination will accept a higher request rate.
| Input | Why it matters |
|---|---|
| Requests per minute | Defines the required completed work. |
| Response size | Affects bandwidth and transfer duration. |
| Measured request duration | Connects throughput with in-flight work. |
| Independent sessions | May require separate allocations even at low throughput. |
| Rotation behavior | Can interrupt traffic or require a pause. |
| Headroom | Allows for observed variability; it is an assumption to validate. |
Measure sustainable throughput on one port
Begin with one connection to a controlled target. Increase concurrency modestly while recording successful completions, elapsed time and failure classes. Stop increasing the rate when the destination's allowed limit, latency requirement or acceptable error behavior is reached.
Use the sustainable result rather than the fastest short burst. Repeat at different times if the workload depends on mobile-network performance. Record the test date, client, target, payload and rotation settings so another person can reproduce the measurement.
Download the capacity worksheetCalculate a provisional port count
For independent work, a starting estimate is required throughput divided by measured sustainable throughput per port, rounded up. Then account for session isolation and the headroom your measurements justify. This arithmetic is a planning estimate, not a performance commitment.
Do not add an arbitrary percentage and label the result guaranteed capacity. Write down why the headroom exists: slower observed periods, a planned rotation pause or a spare allocation for recovery. Validate the proposed configuration with the real workload before purchasing a large fleet.
throughput_ports = ceil(required_completed_requests_per_minute
/ measured_sustainable_requests_per_port)
provisional_ports = max(throughput_ports, independent_port_sessions)
+ explicitly_planned_spare_ports
Monthly cost estimate = provisional_ports × selected_port_priceIf measured sustainable throughput is zero or unknown, the calculation is not meaningful. Complete the one-port test first.
Account for sessions and rotation
Stable connection credentials and a stable public exit IP are different requirements. A dedicated modem can still receive a different carrier address. If a workflow cannot tolerate an address change, establish that requirement before selecting a mobile plan.
A spare port may help recover a failed task, but it does not preserve the same carrier IP or browser session automatically. The number of spare devices alone cannot justify a claim such as 99.99% uptime. That requires a defined service boundary and actual evidence.
Connect the capacity estimate to the budget
Use the cost calculator to compare metered bandwidth with the number of dedicated ports your workload needs. Keep the currency, rental period, taxes and extra fees consistent. A lower arithmetic total is not a useful saving if the selected plan cannot meet the workload's session or throughput requirements.
Start with a small validation purchase and use its results to revise the estimate. Check the current catalog and available inventory at checkout rather than treating a historical price or country page as a reservation.
Sources and further reading
The guidance above distinguishes documented client behavior from product-specific details. Confirm current plan availability at checkout.