Glossary
Short definitions of the terms used in these docs and in the UI. In the Turkish pages we keep some engineering terms in English on purpose (runner, trace, span, executor): teams say them that way at work, and the Turkish words blur the meaning. Settled Turkish words (load test, run, step, scenario, threshold, latency, connection) stay Turkish there.
An example: a 300-VU run is split across 3 runners, each running 100 virtual users; when p95 latency crosses the 500 ms threshold, the run becomes "Thresholds failed".
| Concept | Meaning |
|---|---|
| Controller | The central service that serves the web UI, the REST API and the gRPC endpoint runners connect to. Keeps tests, connections, users and results in Postgres; splits the load across runners. |
| Runner | The process (machine) that generates the load. Connects to the controller with a key and sends metrics every second. With several runners, the VUs and the rate are split evenly (300 VUs, 3 runners → 100 VUs each). The license limits the runners in one run. |
| Location | Where a runner sends load from (e.g. istanbul, frankfurt); given when the runner is installed. A run can be split across locations by percentage (e.g. 60% / 40%); a location's share is spread evenly over its runners. The license limits the locations in one run. |
| Test | A saved load test definition: scenarios, thresholds, variables, environments. Every save is a new version. |
| Scenario | A user flow inside a test with its own executor and step list. Several scenarios of a test can run at the same time. |
| Step | One request or operation: an HTTP request, a SQL query, a Kafka produce… |
| Executor / load model | The shape of the load: how many VUs, at what rate, for how long (constant-vus, ramping-arrival-rate…). |
| VU (virtual user) | An independent loop that runs the scenario's steps from start to end, again and again. It waits for each answer, so requests per second follow from the VU count and the server's speed. 100 VUs means 100 people at once. |
| Iteration | One VU running the scenario once from start to end. With 3 requests in the scenario, 1 iteration is 3 requests; steps that run once at the start (like a sign-in) are not repeated. |
| Iterations/s | How many iterations start each second (arrival-rate executors and the breaking point test); not a VU count. 10 iterations/s × 3 requests = 30 requests a second. Spitfire runs the VUs needed. |
| Ramp-up | The phase in which the VU count or the request rate grows over time. |
| Think time | A pause between two steps that imitates a real user. |
| Check | Whether a response is right (status, JSONPath, XPath, regex, header, latency). A failed check counts as an error. |
| Extract | Taking a value (e.g. a token) from a response into a variable used later as {{variable}}. |
| Threshold | The run's pass/fail verdict, e.g. req_duration p(95)<500. A crossed threshold makes the run "Thresholds failed"; optionally it stops the run. |
| Run | One execution of a test and its results. |
| Baseline | The reference run later runs are compared with. |
| Connection | The address and credentials of a database, broker or gRPC service; secrets are stored encrypted. |
| Secret | A password, token, client secret…; never shown again in the UI, only replaced. |
| Data file | A CSV that feeds values into steps row by row. |
| Workspace | On the Agency plan, each client's separate area, invisible to the others. |
| Group | A list that decides which tests users may see and run. |
| API token | A personal key CI pipelines use to run tests on your behalf. |
| Setup code | The one-time code the installer prints for creating the first admin. |
| License key | A signed key starting with SPF1.…, verified without going online. |
| Audit log | Who did what, and when; an append-only record. |
| Breaking point | The load level where thresholds are first crossed or the error rate shoots up; a breaking-point run searches for it step by step. |
| Error rate | Failed requests (connection error, unexpected status, failed check) over all requests. |
| Trace and span | A request's journey across services (trace) and each operation in it (span): an HTTP call, an SQL query. |
| Backend tab | The run page tab that lines up APM data with the load steps: which service or query slowed down at which step. |