System Requirements

CPU: 2 cores (2.4GHz+)
RAM: 4GB
Storage: 20GB SSD
Network: 100 Mbps

Expected Capacity:
- Concurrent Users: 50-100
- Daily Submissions: 500
- Storage Growth: ~1GB/month

CPU: 4-8 cores (2.4GHz+)
RAM: 8-16GB
Storage: 100GB SSD
Network: 1 Gbps

Expected Capacity:
- Concurrent Users: 200-500
- Daily Submissions: 2,500
- Storage Growth: ~5GB/month

CPU: 16+ cores (2.4GHz+)
RAM: 32GB+
Storage: 500GB+ SSD (RAID recommended)
Network: 1-10 Gbps

Expected Capacity:
- Concurrent Users: 1,000+
- Daily Submissions: 10,000+
- Storage Growth: ~20GB/month

Supported Distributions:
  Ubuntu: 20.04 LTS, 22.04 LTS
  Debian: 11, 12
  RHEL/CentOS: 8, 9
  Amazon Linux: 2, 2023

Required Packages:
  - python3 (3.8+)
  - python3-pip
  - python3-venv
  - git
  - sqlite3
  - nginx (recommended)
  - supervisor/systemd

Versions: 11.0+ (Big Sur or later)

Required Tools:
  - Homebrew
  - Python 3.8+ (via brew)
  - Git
  - SQLite (included)

Versions: Windows 10, Windows Server 2019+

Required Software:
  - Python 3.8+ (python.org)
  - Git for Windows
  - Visual C++ Build Tools
  - Windows Terminal (recommended)

Note: WSL2 recommended for production

Python Version: 3.8 minimum, 3.10+ recommended

Core Dependencies:
  - fasthtml: Web framework
  - uvicorn: ASGI server
  - sqlite3: Database (included)
  - bcrypt: Password hashing
  - python-multipart: File uploads
  - litellm: AI provider integration
  - cryptography: Encryption

Development Tools:
  - pytest: Testing
  - black: Code formatting
  - mypy: Type checking
  - pip-tools: Dependency management

Version: 3.32.0+ (included with Python)

Configuration:
  - WAL mode enabled
  - Foreign keys enabled
  - Auto-vacuum enabled

Limitations:
  - Single file database
  - Limited concurrent writes
  - Max size: ~280TB (practical: 10GB)

Version: 12+

Benefits:
  - Better concurrency
  - Advanced features
  - Horizontal scaling

Requirements:
  - Dedicated database server
  - Connection pooling
  - Regular maintenance

Server: Uvicorn (included)

Usage:
  - Development only
  - Single process
  - Auto-reload

Command: python app.py

Reverse Proxy: nginx

Configuration:
  - SSL termination
  - Static file serving
  - Load balancing
  - Rate limiting

Application Server: Uvicorn + Gunicorn

Configuration:
  - Multiple workers
  - Process management
  - Graceful reloads
  - Health checks

Minimum Bandwidth:
  Per User: 1 Mbps
  API Calls: 10 Mbps reserved

Recommended:
  Institution: 100 Mbps+
  Peak Usage: 500 Mbps

Considerations:
  - File uploads (10MB max)
  - AI API calls
  - Concurrent users
  - Backup transfers

Required Ports:
  HTTP: 80 (redirect to HTTPS)
  HTTPS: 443 (main access)

Internal Ports:
  Application: 5001 (default)
  Database: N/A (file-based)

Outbound Connections:
  HTTPS: 443 (AI APIs)
  SMTP: 587/465 (email)

Domain Setup:
  A Record: feedforward.institution.edu

Optional:
  CNAME: www.feedforward.institution.edu

SSL Certificate:
  Type: Valid CA-signed
  Coverage: Main domain + www
  Renewal: Auto-renewal recommended

OpenAI:
  - API Key required
  - Billing enabled
  - Usage limits configured

Anthropic:
  - API Key required
  - Approved account

Google AI:
  - API Key or Service Account
  - Project configured

Ollama (Local):
  - Local server running
  - Models downloaded
  - Sufficient GPU/RAM

SMTP Requirements:
  - SMTP server access
  - Authentication credentials
  - TLS/SSL support
  - Appropriate send limits

Options:
  - Institution SMTP
  - Gmail (with app password)
  - SendGrid
  - Amazon SES
  - Mailgun

Recommended Services:
  - Application monitoring (New Relic, DataDog)
  - Log aggregation (ELK, Splunk)
  - Uptime monitoring (Pingdom, UptimeRobot)
  - Error tracking (Sentry)

Requirements:
  - Valid SSL certificate
  - TLS 1.2 minimum
  - Strong cipher suites
  - HSTS enabled

Certificate Options:
  - Let's Encrypt (free)
  - Commercial CA
  - Institution wildcard

Inbound Rules:
  - 443/tcp from anywhere (HTTPS)
  - 80/tcp from anywhere (redirect)
  - 22/tcp from admin IPs (SSH)

Outbound Rules:
  - 443/tcp to anywhere (APIs)
  - 587/tcp to SMTP server
  - 53/udp to DNS servers

Internal:
  - 5001/tcp localhost only

Server Access:
  - SSH key authentication
  - Sudo access limited
  - Regular security updates

Application Access:
  - Strong admin passwords
  - Regular key rotation
  - Audit logging enabled

CPU Scaling:
  - 1 core per 50 concurrent users
  - Additional cores for AI processing

RAM Scaling:
  - 100MB per concurrent user
  - 2GB base system requirement
  - Cache allocation extra

Storage Scaling:
  - 10MB per submission average
  - 1GB per 1000 users
  - 20% overhead for indexes

Storage Needs:
  - 3x database size minimum
  - Remote backup location
  - Encryption capability

Backup Frequency:
  - Database: Daily
  - Configuration: On change
  - Logs: Weekly rotation

RTO (Recovery Time Objective): 4 hours
RPO (Recovery Point Objective): 24 hours

Test Schedule:
  - Monthly restore test
  - Quarterly DR drill
  - Annual full recovery

Considerations:
  - Data must stay in country/region
  - AI API calls may cross borders
  - Backup location compliance

Solutions:
  - Local deployment
  - Regional AI endpoints
  - Compliant backup services

Logging:
  - All authentication attempts
  - Data access patterns
  - Configuration changes
  - API usage

Retention:
  - Audit logs: 1 year
  - Access logs: 90 days
  - Error logs: 30 days

Purpose: Local development

Requirements:
  - Python 3.8+
  - Git
  - Text editor/IDE
  - 2GB RAM free
  - 5GB disk space

Optional:
  - Docker Desktop
  - PostgreSQL (testing)
  - Multiple browsers

Purpose: Pre-production testing

Requirements:
  - Mirrors production
  - Separate database
  - Test data only
  - Same OS version
  - Similar resources (50%)

Purpose: Live system

Requirements:
  - Meets all requirements above
  - Redundant components
  - Monitoring active
  - Backup configured
  - Security hardened