Performance Testing Results
The goal of this test was to benchmark the Decision Engine service and establish expected performance metrics under controlled conditions.
-
Load: 160 requests per second (RPS)
-
Duration: 10 minutes
-
Total Requests Sent: 960,000
-
CPU Target: Approximately 1 core usage
-
Environment: Standardised hardware and deployment configuration.
-
Throughput: ~159.34 requests per second
-
Total Requests Processed: 956,040
-
CPU Usage: ~1 core consistently
-
Latency Metrics:
- TP90 (90th percentile): 0.263 seconds
- TP95 (95th percentile): 0.366 seconds
- TP99 (99th percentile): 0.785 seconds
The test successfully validated that Open-Router can maintain reliable performance when operating at its targeted capacity of one CPU core. By sustaining a constant load of 160 requests per second (RPS) for 10 minutes, a total of 960,000 requests were sent.
The system demonstrated predictable and stable behavior, with actual throughput averaging 159.34 RPS, closely matching the target. This minimal variance highlights the system's efficiency with negligible overhead or performance degradation.
Testing methodology ensured accuracy by:
- Maintaining independent request generation regardless of response times
- Using concurrent execution to support overlapping requests
- Implementing fixed-interval scheduling to maintain consistent RPS
Response time analysis showed that even under continuous load, the system maintained fast processing speeds.
Based on the above results, the Decision Engine service can be expected to handle approximately 160 RPS sustainably using ~1 CPU core, while maintaining latency characteristics suitable for high-throughput transaction routing systems.