performance-benchmark-analysis

To validate the efficiency and scalability of our proposed architecture, we conducted a series of rigorous performance tests within the GNS3 simulation environment. The goal was to prove that the “Scout-Killer” model provides superior security without degrading host performance.

1. Agent Resource Footprint (CPU/RAM)

One of the critical requirements was minimal impact on the client nodes. We measured the resource consumption of the LocalAgent over a 24-hour period.

• Average CPU Load: Consistent at 0.08% - 0.12% on a single-core vCPU.
• Memory Usage: Java Heap stabilized at 24MB after JIT compilation.

Verification Command: To demonstrate this low overhead, we used pidstat to isolate the Java Agent process:

# Monitor the LocalAgent process every 1 second
pidstat -r -u -p $(pgrep -f LocalAgent) 1 10

2. Mobility Latency (Scout Deployment)

We measured the “Time-to-Insight”—the duration from anomaly detection on the Central Server to the receipt of the forensic report from the Scout Agent.

• Migration Time (Server to Client): 180ms (avg).
• Execution Time (Process Audit): 45ms.
• Return Trip: 175ms.
• Total Round-Trip Time (RTT): ~400ms.

This sub-second response time proves that JADE’s Weak Mobility is significantly faster than establishing new SSH sessions for every command, especially under network load.

3. Bandwidth Usage & WireGuard Overhead

We analyzed the network impact of wrapping JADE ACL messages inside the WireGuard VPN tunnel.

Conclusion: The overhead introduced by WireGuard (~12%) is negligible for modern broadband networks. The slight increase in packet size is a worthy trade-off for the encryption provided by the ChaCha20-Poly1305 cipher.