This study presents a comparative performance analysis of response times between Serverless architecture (AWS Lambda container-based) and Edge Computing (V8 Isolate-based). The evaluation focused on handling CPU-intensive computational loads using the Hono.js framework via the execution of 1,000 concurrent requests per platform. Measurements were strictly centered on Client-Side Round Trip Time (RTT) to ensure a fair comparison, deliberately bypassing server-side timing due to time quantization security restrictions inherent in the V8 Isolate environment. Empirical results demonstrate that Vercel recorded an average RTT of 432.40 ms, which is 76.29% faster than Cloudflare Workers' average of 762.29 ms. The Mann-Whitney U significance test statistically confirmed this difference (p < 0.05). Furthermore, consistency analysis via the Coefficient of Variation (CV) placed Vercel ahead with a highly stable value of 0.116 compared to Cloudflare's 0.527. Cloudflare also exhibited a larger cold start latency gap during the warm-up phase (28.03 ms difference) compared to Vercel (7.16 ms difference). The conclusion indicates that for REST API applications with heavy computational loads, container-based serverless architecture provides significantly superior stability, lower tail latency, and overall efficiency compared to Edge Isolate, challenging the theoretical assumption of absolute Edge network superiority in industrial scenarios.
Copyrights © 2026