Server Architecture

​

Overview

The Speckle server follows a microservices architecture pattern, with each component responsible for specific functionality while communicating through well-defined APIs. This design allows for:

• Scalability: Components can be scaled independently
• Reliability: Failure in one component doesn’t bring down the entire system
• Maintainability: Each component can be updated and maintained separately
• Flexibility: Components can be deployed on different infrastructure as needed

​

Core Services

​

API Server

The central component that handles all client requests and business logic.

Responsibilities:

• Authentication & Authorization
• Project Management
• Data Operations
• User Management
• Webhook Management

​

Frontend

The web interface for managing Speckle projects and data.

Responsibilities:

• User Interface
• Data Visualization
• User Management
• Project Administration

​

Supporting Services

• Preview Service: Generates object previews headlessly
• Webhook Service: Handles external webhook calls and integrations
• File Import Service: Parses and imports various file formats
• Database monitor: Generates statistics related to the stored data

​

Required Dependencies

​

Database (PostgreSQL)

PostgreSQL serves as the primary database for all relational data including user accounts, projects, object metadata, and activity logs.

​

Object Storage (S3-compatible)

Object storage handles the actual 3D data and file assets. We require an S3-compatible blob storage, such as AWS S3 or MinIO.

​

Cache Layer (Redis-compatible)

The cache layer provides caching and session management capabilities, serving as the Redis-compatible open-source alternative.

Cached Data:

• User Sessions
• Rate Limiting
• Messages between supporting services and the API server

​

Email server (SMTP-compatible)

This handles all transactional emails to users, including account email verification. We recommend using an externally provided email service which is SMTP compatible.