Server Overview
Contents
Server Overview
Spire is a distributed trading platform composed of multiple specialized servers that work together to provide a complete trading infrastructure. Each server is independently deployable, communicates through well-defined service interfaces, and authenticates through a central Service Locator.
Architecture Principles
Service-Oriented Design - Each server provides a focused set of functionality
Distributed Deployment - Services can be deployed across multiple machines for scalability
Centralized Authentication - All services authenticate through the Service Locator
Persistent Storage - Critical data persisted to MySQL for durability
Real-time Updates - Services publish updates to subscribed clients
Service Categories
Infrastructure Services
Service Locator - Centralized authentication, account management, and service discovery
Uid Server - Generates unique identifiers for orders and other entities
Account & Administration
Administration Server - Manages account roles, entitlements, risk parameters, and trading permissions
Web Portal - Web-based interface for account management and administrative tasks
Market Data
Definitions Server - Provides reference data (countries, currencies, venues, trading schedules)
Market Data Server - Receives, sequences, and persists market data from feed clients
Market Data Relay Server - Load-balanced distribution of market data to end users
Charting Server - Processes time series queries for candlestick and technical analysis data
Trading Controls
Compliance Server - Manages compliance rules and monitors violations
Risk Server - Monitors profit/loss, enforces risk limits, manages account state transitions
Order Execution Server - Routes orders to venues, tracks execution, enforces controls
Common Operational Patterns
All servers follow consistent operational patterns:
Configuration via YAML files
Setup scripts for generating configurations
Standard start/stop/check scripts
Timestamped log files in logs/ directory
PID-based process management
System-Wide Management
In addition to individual server management scripts, the platform provides aggregate scripts for managing all servers collectively.
install.sh
The install.sh script is designed for the latest Ubuntu Server LTS and performs a complete system installation including:
- Installing system dependencies (build tools, MySQL, Node.js, Python packages)
- Building all server applications
- Configuring MySQL database and creating the spire schema
- Creating service accounts in the Service Locator
- Setting up initial permissions and directory structure
- Configuring all servers with appropriate network settings
Usage:
./install.sh -p [password] [options] Required: -p Password for Spire services Optional: -u MySQL username (default: spireadmin) -m MySQL password (default: same as -p) -i Global network interface (default: auto-detected)
The installation script automatically creates service accounts for all servers and assigns them to the appropriate permission groups.
setup.py
The setup.py script configures all servers by propagating common settings across individual server configurations:
python setup.py [options] Optional: --local Local network interface --world Global/public network interface --address Service Locator address --password Service account password --mysql_address MySQL server address --mysql_username MySQL username --mysql_password MySQL password --mysql_schema MySQL schema name
This script calls each server's individual setup script with appropriate parameters, ensuring consistent configuration across the platform.
start.sh
The start.sh script starts all servers in dependency order:
- Service Locator
- Uid Server
- Definitions Server
- Administration Server
- Market Data Server
- Market Data Relay Server
- Charting Server
- Compliance Server
- Order Execution Server
- Risk Server
- Web Portal
- Market Data Feed Client
After starting all servers, the script automatically runs reset_risk_states.py to initialize risk state for the trading session.
stop.sh
The stop.sh script gracefully stops all servers in reverse dependency order, ensuring that dependent services shut down before the services they depend on.
check.sh
The check.sh script verifies that all servers are running and reports the status of each service. It exits with a non-zero status if any server is not running, making it suitable for monitoring and health check automation.