Proxy quality is three tests: speed, stability, and IP freshness, measured on the actual workload rather than on a generic benchmark. A proxy gateway pool is only as good as these three numbers, and the test procedure that measures them consistently is what separates a managed pool from a guess.
This guide explains the three tests, the procedure that runs them, and what good looks like for a proxy pool.
The Three Tests
The three tests answer the three questions a workload asks: is the path fast enough, does it stay stable through the session, and are the IPs fresh enough for the task? Each test measures a different property, and a pool can pass one and fail another.
Speed is measured as latency and throughput on the egress path: the round-trip time for a request and the bandwidth available for the transfer. The measurement must reflect the workload, because a path that is fast for small requests can be slow for large ones.
Stability is measured across time: how long the session holds, how often the address or path changes mid-session, and how the connection behaves under load. IP freshness is measured by the addresses' history and how the target services treat them. The three tests together define the pool's quality.
Speed
The speed test measures the egress path under realistic conditions: latency to the target, throughput for the transfer size, and the variance across the day. The measurement is repeated on the same paths at different hours, because carrier and network conditions change with load.
The test should use the actual workload: a data-collection task requests small pages, while a transfer task moves large files, and the throughput number that matters is the one the task experiences. A generic speed test flatters paths that the workload would never use that way.
The speed record includes the context: the path, the carrier, the time, and the measured numbers, because the record is what makes the next measurement comparable. Without the context, a speed number is an anecdote.
The variance across the day is often the more useful number than the average: a path with a great average and a bad evening peak is a path that fails when the workload runs. The distribution is the number to manage.
| Metric | What it measures | What good looks like |
|---|---|---|
| Latency | Round-trip time | Matches the workload's needs |
| Throughput | Transfer rate | Sufficient for the task |
| Variance | Stability across the day | No dropouts at peak |
| Error rate | Failed requests | Near zero on healthy paths |
The table is the speed test in one view: each metric has a definition and a target, and the targets come from the workload, not from a universal number.
Stability
The stability test checks the session behavior: how long a connection holds its path, how often the address changes mid-session, and how the pool behaves under load. A workload that needs persistence depends on this test, and the result decides the rotation mode.
The test runs sessions of the length the workload uses, through each pool, and records the address changes and the disconnects. The pattern, not a single session, is the useful output, because a path that drops once in ten sessions is different from one that drops every time.
The stability result feeds the configuration: pools that need persistence use static or session-based rotation, and pools that tolerate change use rotation. The test is what matches the mode to the workload.
The stability test also covers the gateway side: how the pool behaves during a SIM change, a carrier degradation, or a gateway reboot, because the pool's resilience is part of its stability. The failure-mode test is what the daily sessions cannot show.
IP Freshness
The IP freshness test checks what the target services see: whether the addresses are recognized, flagged, or blocked, and whether the pool's addresses appear fresh to the workload's targets. The test is workload-specific, because the same address can be treated differently by different services.
The measurement is a sample: run the workload's requests through the pool, observe the target's response, and record the address and the outcome. The pattern of blocks and flags across the pool is the freshness score.
Freshness is a carrier and rotation property: fresh addressing comes from the carrier's allocation and the pool's rotation, and it degrades with heavy shared use. The freshness test is repeated on a schedule, because the property changes as the fleet and the targets change.
The freshness record should include the target's behavior over time: which addresses were accepted, which were challenged, and which were blocked, because the pattern reveals whether the issue is the address pool or the workload's request pattern. The record turns blocking from a surprise into a trend.
The Test Procedure
The procedure is a scheduled loop: define the workload's metrics, run the three tests on each pool at fixed times, record the results with context, and review the record weekly. The loop turns quality from a feeling into a managed number.
The procedure includes the baseline: the first two weeks of measurements set the normal range per pool, and later results are compared against it. The baseline is what makes a regression visible, and it is refreshed quarterly.
The procedure ends with the action: a pool that misses a metric gets a configuration change, a SIM check, or a re-test, and the action is recorded with the result. The loop is what keeps the pool quality from drifting.
The procedure should be documented so any operator can run it: the commands or steps for each test, the recording format, and the review schedule, because a procedure that lives in one person's head dies with their departure. The documentation is part of the procedure.
What Good Looks Like
Good looks like a pool whose metrics match the workload: latency within the task's tolerance, throughput sufficient for the transfer, sessions holding for the needed duration, and addresses that the targets accept. The specific numbers come from the workload, not from a generic ideal.
Good also looks like consistency: the same metrics across the day, low variance on healthy paths, and a record that shows the pool improving or holding rather than degrading. The consistency is the operational signature of a managed pool.
Finally, good looks like recoverability: when a path degrades, the pool routes around it, the record shows the event, and the review produces a fix. The recoverability is what makes the quality durable.
Good also looks like a connection between the tests and the workload outcome: the pool's metrics should explain why a task succeeded or failed, and the review should connect the two. The connection is what makes the testing worth doing.
Telarvo Expert Views
The three tests are only useful when they measure the real workload and run on a schedule: a speed test against a generic site tells you little about the pool the task uses. Define the metrics from the task, baseline the pool, and review the record weekly.
— Network Infrastructure Engineer, Telarvo Store
Validation note: target behavior and network conditions change; re-test on a schedule and compare against your own baseline.
Conclusion
Proxy quality is speed, stability, and IP freshness measured on the real workload, run on a scheduled procedure with a baseline, and reviewed weekly so the pool stays managed.
Key Takeaways for B2B Buyers
Measure the three tests on the actual workload, record the results with context, set the baseline from the first two weeks, and review the record weekly with an action for each miss.
Questions to Ask Before Committing
Ask what the gateway reports per path, how pools are configured for rotation, what the monitoring covers, and how the supplier documents quality testing.
Ask Telarvo Store which proxy gateway configuration supports your quality-testing routine before you deploy.
FAQs
How often should I test proxy quality?
Run the scheduled loop continuously or weekly, with a baseline from the first two weeks and a re-test after any change.
What is a good proxy latency?
It depends on the workload and the market; measure the task's tolerance and compare the pool against its own baseline.
Why do my IPs get blocked?
Freshness depends on the carrier's allocation, the pool's rotation, and the targets' behavior; the freshness test finds the pattern.
Does rotation improve stability?
No; rotation changes addresses, which helps freshness and hurts persistence. Choose the mode by the workload's need.
Can I test proxy quality without special tools?
Basic tests use common tools for latency and throughput, and the workload itself is the freshness test; the key is recording the results consistently.