How open-eo/openeo-geopyspark-driver works
Main Components
a. Client Layer:
• REST API Interface (Flask-based)
• Direct API Interface (non-REST)
• OpenEO API endpoints
b. Core Processing Layer:
• GeoPySpark/Geotrellis Engine
• Spark Processing Framework
• Batch Job Processing System
• User-Defined Functions (UDF) Handler
c. Storage Layer:
• Accumulo Tile Storage
• Object Storage
• GeoTiff File Storage
• STAC Catalog Integration
d. Infrastructure Components:
• Kubernetes/YARN Cluster
• Zookeeper (for service discovery)
• NginX Reverse Proxy
• Prometheus (monitoring)
• Traefik (routing)
Component Relationships
• Client requests → REST/Direct API
• API Layer → Process Graph Executor
• Process Graph Executor → GeoPySpark Engine
• GeoPySpark Engine → Spark Processing
• Spark Processing ↔ Storage Systems
• Infrastructure Services ↔ All Components
Specific Implementation Details
Specialized components:
• Batch Job Processing System
• Async Task Handling
• ML Model Integration (CatBoost, Random Forest)
• Vector Cube Processing
• STAC API Integration
• ETL Pipeline
Deployment Context
Deployment options:
• Kubernetes Deployment
• YARN Cluster Deployment
• Local Development Setup
• Docker Container Environment
Key Integration Points
Integration with:
• Sentinel Hub
• CREO/OSCARS Catalogs
• External Authentication Systems
• Monitoring Systems