Архитектура
Руководство по интеграции
Интеграция с quic-test и CloudBridge Relay
AI Routing Lab - Integration Guide
How to integrate AI Routing Lab with quic-test and CloudBridge Relay
Integration Overview
AI Routing Lab integrates with two main components:
- quic-test - For data collection and model validation
- CloudBridge Relay - For production route optimization
Integration with quic-test
Step 1: Setup quic-test with Prometheus Export
# Start quic-test server with Prometheus metrics
cd cloudbridge/quic-test
./bin/quic-server --prometheus-port 9090 --prometheus-path /metrics
Step 2: Configure Prometheus
# prometheus.yml
scrape_configs:
- job_name: 'quic-test'
scrape_interval: 15s
static_configs:
- targets: ['localhost:9090']
Step 3: Collect Data
from data.collectors.quic_test_collector import PrometheusCollector
collector = PrometheusCollector(prometheus_url="http://localhost:9090")
metrics = collector.collect_all_metrics()
Step 4: Validate Predictions
# Run quic-test with ML validation mode
./bin/quic-client \
--ml-validation-mode \
--ml-predictions-file predictions.json \
--output results.json
Integration with CloudBridge Relay
Step 1: Deploy ML Prediction API
# Start inference service
python inference/predictor.py --model models/latency_predictor.pkl --port 8080
Step 2: Configure CloudBridge Relay
# relay-config.yaml
ml_routing:
enabled: true
prediction_api_url: "http://ai-routing-lab:8080/api/v1/predict"
fallback_to_baseline: true
prediction_timeout: 100ms
Step 3: Route Selection
CloudBridge Relay will query ML API for route predictions and select optimal route based on predictions.
Data Collection Workflow
- quic-test generates traffic → Exports metrics to Prometheus
- AI Routing Lab collects metrics → Processes and stores
- ML models train → Generate predictions
- Predictions validated → Against quic-test results
- Production deployment → CloudBridge Relay uses predictions
Validation Workflow
- Generate predictions for test routes
- Run quic-test with same routes
- Compare predictions vs actual latency/jitter
- Calculate accuracy metrics
- Report results and iterate
For detailed integration code, see: integration/quic_test_client.py and integration/relay_integration.py