Running a cloud platform means constantly balancing performance and cost. In this document, we'll break down the infrastructure expenses for TRAST, explaining how we've structured our systems to maxim
Executive summary
This document outlines the comprehensive infrastructure and operational costs for TRAST platform. Our cost structure is optimized for scalability, reliability, and performance while maintaining cost-effectiveness.
Cloud infrastructure
The TRAST platform is built for scalability and reliability. The primary database is a managed PostgreSQL instance on Supabase, with egress (0/250GB per month) and Monthly Active Users (0/100,000 MAU) quotas monitored for cost and performance. The bot is deployed on Heroku with auto-scaling enabled, ensuring seamless scaling and high availability. All infrastructure and backend services are designed to scale horizontally as user demand increases.
Compute resources (Heroku)
Production environment
• Heroku dynos: auto-scaling enabled
- High availability setup
- Multi-region deployment (if needed)
• Supabase PostgreSQL: managed database
- Real-time data access
- Automated backups
- Egress and MAU quotas monitored
Development/staging
• test environments: $400/month
• CI/CD pipeline: $300/month
Language models
• openai GPT-4: $2,000/month
- Managed via LLMService
- Token optimization
- Caching strategy
- Rate limiting
Custom models and future providers can be integrated as needed.
NOTE: All costs are estimates and May vary based on actual usage patterns and market conditions. Regular reviews and adjustments will be performed to maintain optimal cost-efficiency.
