Summary

WASM SDK decryption performance is significantly slower than the JS SDK, taking 466.5ms compared to JS SDK's 202.7ms (2.3x slower).

Benchmark Environment

• Platform: macOS (Apple Silicon)
• Browser: Chromium 141.0.7390.37
• Test Message: 39 bytes
• Network: localhost (minimal latency)
• Test Framework: Playwright E2E

Performance Metrics

Analysis

The WASM decryption includes:

• Full KAS rewrap protocol
• RSA-OAEP key unwrapping
• JWT signing with ES256
• AES-256-GCM decryption

Suspected bottlenecks:

1. JWT signing operations - May not be optimized
2. RSA key operations - Potentially using slower algorithms
3. Network requests - Possibly not reusing connections
4. Token caching - May not be caching KAS responses

Expected Outcome

WASM should match or exceed JS SDK performance for decryption, ideally:

• Target: <200ms for 39-byte payload
• Stretch goal: <150ms (faster than JS SDK)

Suggested Optimizations

1. Connection Pooling: Reuse HTTP connections to KAS
2. Token Caching: Cache KAS rewrap responses where possible
3. Crypto Optimization: Use browser's WebCrypto API via wasm-bindgen where appropriate
4. JWT Optimization: Optimize ES256 signing performance
5. Async Operations: Ensure all I/O is truly non-blocking

Test Repository

Full benchmark available at: wasm-vs-js comparison app