Everything you need to know before running tests.
LatencyRadar measures how long your API or website takes to respond from multiple global regions.
A free snapshot runs a single request per region.
Full analysis runs repeated samples to calculate p50, p95 and p99 latency, plus protocol breakdown (DNS, TLS, connect, TTFB, download).
No.
This is a distributed latency diagnostic tool, not a stress-testing platform.
We send a limited number of requests per region and strictly cap concurrency. It is designed to measure performance - not to simulate traffic spikes.
No.
The platform is intentionally designed to prevent abuse.
No.
You can run a free snapshot immediately.
Full analysis is a one-time unlock per test and does not require account creation.
We store:
We do not store:
Test data is retained for a limited period for operational purposes.
Yes.
You may provide custom headers (for example Authorization headers).
However, you are responsible for ensuring you are authorized to test the endpoint.
Latency is measured from real VPS servers located in multiple regions.
Full analysis uses repeated sampling to compute statistically meaningful percentiles (p50, p95, p99).
Results reflect real-world network conditions at the time of testing.
Network distance, routing, DNS resolution, TLS handshake time, and server location all affect latency.
Your local environment may not represent global performance.
p95 means that 95% of requests completed within that time.
It is commonly used in performance engineering because it reflects worst-case user experience better than average latency.
No.
This tool is provided as a diagnostic utility without uptime guarantees or service-level agreements.
Due to the nature of digital diagnostic reports, purchases are generally non-refundable.
If a test failed due to an internal platform error, please contact support.