Benchmark Lab — Aether Search

Benchmark Lab fires real search requests against the live server and measures throughput and latency. Choose Single Concurrency for a fixed load level, or Latency Growth to sweep through a range of concurrency values and see how the system scales.

Test Configuration

Benchmark Parameters

Single: fires N requests at a fixed concurrency and measures average, P99, and RPS. Latency Growth: sweeps from min to max concurrency to plot how latency scales under load — ideal for capacity planning.

Total Requests

Total number of search requests to fire across all workers

Concurrency Level parallel workers

How many requests run simultaneously

Query Settings

Search Endpoint Type

Typo correction is slightly heavier — a good test of worst-case latency

Search Query

All benchmark requests use this fixed query string

Benchmark API Key

Required to authenticate benchmark requests

Results

Scenario Metrics — Final Averages

RPS = requests per second (throughput). Avg Latency = mean round-trip time. P99 = the slowest 1% of requests — use this for SLA planning. Higher concurrency generally increases RPS up to a point, after which latency grows.

Throughput

requests / sec

Average Latency

0.00

milliseconds

P99 Latency

0.00

milliseconds

Successful Requests

0 / 0

success / total

Latency Distribution & Percentiles

The chart shows latency percentile distribution across the benchmark run. A tight distribution indicates consistent performance.