Fast Mermaid Validator MCP

# Mermaid Validator API - Performance Testing & Load Limits ## Apache Bench Load Testing Results ### Test Environment - **Tool**: Apache Bench (ab) Version 2.3 - **Deployment**: mmjc-dev namespace (single pod) - **Test data**: Simple flowchart diagram (453 bytes response) - **Configuration**: v1.0.15 with increased limits ### Performance Degradation Analysis #### Baseline Performance (10 concurrent users) ```bash ab -n 100 -c 10 -T application/json -p test_data.json http://localhost:8086/api/v1/validate ``` - **Requests per second**: 5.83 req/sec - **Response time**: 1714ms mean, 171ms per concurrent request - **Failed requests**: 6/100 (6% failure rate) - **Status**: ✅ Stable performance #### Moderate Load (20 concurrent users) ```bash ab -n 200 -c 20 -T application/json -p test_data.json http://localhost:8086/api/v1/validate ``` - **Requests per second**: 11.65 req/sec (2x improvement) - **Response time**: 1716ms mean, 86ms per concurrent request - **Failed requests**: 6/200 (3% failure rate) - **Status**: ✅ Good performance scaling #### High Load (50 concurrent users) ```bash ab -n 500 -c 50 -T application/json -p test_data.json http://localhost:8086/api/v1/validate ``` - **Requests per second**: 27.10 req/sec (4.6x improvement) - **Response time**: 1845ms mean, 37ms per concurrent request - **Failed requests**: 31/500 (6.2% failure rate) - **Status**: ⚠️ Starting to show stress #### **Degradation Point (100 concurrent users)** ```bash ab -n 1000 -c 100 -T application/json -p test_data.json http://localhost:8086/api/v1/validate ``` - **Requests per second**: 48.75 req/sec (8.3x improvement) - **Response time**: 2051ms mean, 20ms per concurrent request - **Failed requests**: 72/1000 (7.2% failure rate) - **Max response time**: 3171ms (significant degradation) - **Status**: 🔴 **DEGRADATION THRESHOLD REACHED** ## Performance Characteristics ### Single Pod Limits - **Optimal concurrency**: 20-50 concurrent requests - **Degradation starts**: ~50 concurrent requests - **Hard limit**: ~100 concurrent requests (7% failure rate) - **Processing capacity**: ~50 req/sec max throughput ### Response Time Analysis | Concurrency | Mean Response Time | 95th Percentile | Failure Rate | |-------------|-------------------|-----------------|--------------| | 10 | 1714ms | 1793ms | 6% | | 20 | 1716ms | 1812ms | 3% | | 50 | 1845ms | 2050ms | 6.2% | | **100** | **2051ms** | **2749ms** | **7.2%** | ## HPA Configuration for Fast Scaling ### Updated HPA Settings ```yaml # Fast stabilization for both environments behavior: scaleDown: stabilizationWindowSeconds: 60 # 1 minute (was 5 minutes) policies: - type: Percent value: 50 # 50% reduction per cycle (was 10%) periodSeconds: 15 # Every 15 seconds scaleUp: stabilizationWindowSeconds: 0 # Immediate scaling policies: - type: Percent value: 200 # Triple pods if needed (dev) periodSeconds: 10 # Every 10 seconds - type: Pods value: 8 # Add max 8 pods at once (dev) periodSeconds: 10 ``` ### Environment-Specific Scaling #### mmjc-dev (Development) - **Min/Max replicas**: 2-20 pods - **Scale-up**: 200% increase, +8 pods max per 10s - **Scale-down**: 50% reduction per 15s after 1min stabilization - **Target capacity**: 20 × 50 = **1,000 req/sec** #### mmjc-test (Testing) - **Min/Max replicas**: 2-15 pods - **Scale-up**: 150% increase, +5 pods max per 10s - **Scale-down**: 50% reduction per 15s after 1min stabilization - **Target capacity**: 15 × 50 = **750 req/sec** ## Environment Variable Configuration ### Current Production Limits ```bash # File handling MAX_FILES=100000 # 100k files MAX_FILE_SIZE=104857600 # 100MB per file MAX_REQUEST_SIZE=0 # Unlimited request body # Content processing MAX_DIAGRAM_CONTENT_LENGTH=1048576 # 1MB per diagram MAX_DIAGRAMS_PER_FILE=50 # 50 diagrams per file MAX_TOTAL_DIAGRAMS=200 # 200 total diagrams per request # Mermaid engine MERMAID_MAX_TEXT_SIZE=1048576 # 1MB text processing MERMAID_MAX_EDGES=500 # 500 edges max MERMAID_MAX_VERTICES=200 # 200 vertices max # Processing timeouts VALIDATION_TIMEOUT=30000 # 30 seconds MAX_TIMEOUT_MS=60000 # 60 seconds max ``` ### Unlimited Development Mode ```bash # Set any limit to -1 for unlimited processing MAX_FILES=-1 # Unlimited files MAX_FILE_SIZE=-1 # Unlimited file size MAX_DIAGRAM_CONTENT_LENGTH=-1 # Unlimited content MERMAID_MAX_TEXT_SIZE=-1 # Unlimited mermaid processing ``` ## Monitoring & Alerts ### Key Metrics to Watch - **Response time > 2000ms**: Scale up trigger - **Failure rate > 5%**: Performance degradation - **CPU utilization > 70%**: HPA scale trigger - **Memory utilization > 70%**: HPA scale trigger ### Load Testing Commands ```bash # Quick performance check ab -n 100 -c 10 -T application/json -p test_data.json http://api-endpoint/validate # Stress test (degradation point) ab -n 1000 -c 100 -T application/json -p test_data.json http://api-endpoint/validate # Sustained load test ab -n 5000 -c 50 -T application/json -p test_data.json http://api-endpoint/validate ``` ## Recommendations ### Production Deployment 1. **Minimum 2 pods** for availability 2. **Target 50 concurrent requests per pod** for optimal performance 3. **Monitor failure rates** and scale before 100 concurrent/pod 4. **Use fast HPA scaling** (10s intervals) for traffic spikes ### Development Testing 1. **Test with realistic diagram sizes** (1KB+ content) 2. **Validate under sustained load** (5+ minute tests) 3. **Monitor memory usage** during large file uploads 4. **Test unlimited configuration** impacts ### Performance Optimization 1. **File upload limits** balance usability vs resources 2. **Content size limits** prevent memory exhaustion 3. **Concurrent processing** should not exceed 50/pod sustained 4. **HPA thresholds** should trigger before degradation point