From single-machine smoke tests to distributed load across on-premise, AWS, and Naver Cloud — all on top of JMeter. Analysis runs from simple CLI-based monitoring through to full Grafana + Prometheus time-series tracking, scaled to fit the project's stage and budget.
System performance testing has long defaulted to expensive commercial products like HP LoadRunner. The license is heavy, scenario changes are slow, and the analysis tooling stays closed.
We run the same — and at times deeper — performance tests on top of open-source JMeter. Load distributes across on-premise, AWS, or Naver Cloud, and analysis flows into open-source standards like Grafana and Prometheus for visualization and time-series tracking.
Start with simple CLI-based monitoring, scale to high-fidelity analysis when needed. The same pipeline carries everything from quick validation to large-scale load verification.
Scenario design, distributed load infrastructure, and visualization-to-time-series analysis — connected end to end.
Real-traffic-shaped JMeter scenarios drive ramp-up, spike, and endurance tests — all from the same script base.
The same scenario scales from a single machine smoke test to a multi-node distributed load. Load coordination and result collection are automated.
We've operated load infrastructure directly on on-premise, AWS, and Naver Cloud. Load nodes are configured to match each environment's characteristics.
JMeter results land alongside system metrics (CPU, memory, GC, DB) in Prometheus, and Grafana dashboards trace them on a shared timeline. Bottlenecks become obvious at a glance.
Start fast with CLI-based result analysis; deepen into Grafana + Prometheus when the project needs it. The depth is dialed in stages.
We achieve equivalent — sometimes deeper — performance testing without the cost of products like HP LoadRunner. No license overhead, more focus on scenarios and infrastructure.
We designed real-traffic-shaped JMeter load scenarios and identified system limits and bottlenecks. A Grafana + Prometheus time-series view traced GC patterns, DB response times, and throughput together — establishing a reliable operational baseline.
As part of SK broadband's CDN monitoring upgrade, we ran fine-tuned JMeter load scenarios to verify that logs from hundreds of servers reaching the pipeline simultaneously aggregate without loss and produce numerically accurate results. We staged a range of traffic patterns and validated that the collection and aggregation pipeline produced consistent figures under every condition.