Load testing — how was it conducted, what metrics were monitored?

Load testing is usually conducted using specialized tools (e.g., k6, JMeter, Locust) or custom scripts in Go.

Main steps:

• Defining load scenarios that reflect real user actions.
• Gradually increasing the load (ramp-up) to identify the degradation point.
• Running long-duration tests to check stability.

Metrics monitored:

• Response time (latency) — average, p95, p99.
• Throughput — number of requests per second.
• Errors — percentage of failed requests.
• Server resource usage — CPU, memory, network.
• Number of open connections and pool status.

Example: during testing of an HTTP service, ensuring that p99 latency does not exceed 200 ms at a load of 1000 RPS, and errors are less than 0.1%.