Deployment Requirements

Documentation Index

Fetch the complete documentation index at: https://docs.mypraxos.com/llms.txt

Use this file to discover all available pages before exploring further.

​

Overview

​

System Requirements

​

Minimum Specifications

• CPU: 2 cores
• RAM: 4 GB
• Storage: 20 GB
• Network: Stable internet connection

​

Recommended Specifications

• CPU: 4-8 cores
• RAM: 8-16 GB
• Storage: 50-100 GB SSD
• Network: High-bandwidth, low-latency

​

High-Scale Specifications

• CPU: 16+ cores (distributed)
• RAM: 32+ GB (per pod)
• Storage: 500+ GB SSD
• Network: Load balanced, redundant

​

Software Dependencies

​

Required

• Python: 3.11 or higher
• Docker: 20.10 or higher (for containerized deployment)
• Kubernetes: 1.24+ (for orchestration)
• MongoDB: 4.4+ or Azure Cosmos DB

​

Optional

• Redis: 6.0+ (for local queue, alternative to Azure Service Bus)
• NGINX: Load balancing and SSL termination
• Prometheus: Monitoring and metrics
• Grafana: Visualization

​

Cloud Services

​

Azure (Recommended)

• Azure Kubernetes Service (AKS) - Container orchestration
• Azure Cosmos DB - Primary database
• Azure Service Bus - Message queue
• Azure Key Vault - Secrets management
• Azure Container Registry - Docker images
• Azure Storage - Blob storage
• Application Insights - Monitoring

​

AWS (Alternative)

• EKS - Kubernetes
• DocumentDB - MongoDB compatible
• SQS - Message queue
• Secrets Manager - Secrets
• ECR - Container registry
• S3 - Object storage
• CloudWatch - Monitoring

​

GCP (Alternative)

• GKE - Kubernetes
• Cloud Firestore or MongoDB Atlas - Database
• Cloud Tasks - Message queue
• Secret Manager - Secrets
• Container Registry - Images
• Cloud Storage - Object storage
• Cloud Monitoring - Observability

​

Database Requirements

​

MongoDB

• Version: 4.4+
• Storage: 10 GB
• RAM: 4 GB
• Connection pooling: 100 connections

• Version: 6.0+
• Storage: 50+ GB SSD
• RAM: 8+ GB
• Replica set with 3 nodes
• Connection pooling: 500 connections

• users
• conversations
• user_context
• integrations
• tasks
• logs

​

Azure Cosmos DB

• API: MongoDB
• Throughput: 1000 RU/s minimum
• Consistency: Session (default)
• Multi-region: Recommended for HA

​

Message Queue

​

Azure Service Bus

• Tier: Standard or Premium
• Queues: agent_tasks, scheduled_tasks
• Topic/Subscriptions: Optional for pub/sub
• Max message size: 256 KB
• Message TTL: 14 days

​

Redis (Alternative)

• Version: 6.0+
• Memory: 2-8 GB
• Persistence: RDB + AOF
• Cluster mode: For high availability

​

API Keys & Credentials

​

Required

• LLM Provider
  • Portkey API Key
  • OpenAI API Key (primary)
  • Google AI API Key (secondary)
• Database
  • MongoDB connection string
  • Or Azure Cosmos DB endpoint + key
• Message Queue
  • Azure Service Bus connection string
  • Or Redis connection URL

​

Integration-Specific

• Telegram: Bot token
• Discord: Bot token, Application ID
• Slack: Bot token, Signing secret
• Google: Client ID, Client Secret
• Microsoft: Client ID, Client Secret, Tenant ID
• Notion: Integration token
• Trello: API key, Token

​

Networking Requirements

​

Inbound

• Port 8000: HTTP/HTTPS (API)
• Port 443: HTTPS (webhooks)

​

Outbound

• LLM providers (OpenAI, Google)
• Integration APIs (Google, Microsoft, Telegram, etc.)
• Database servers
• Message queue services

​

Webhooks

• Telegram webhooks
• Discord webhooks
• Slack events
• Google Calendar notifications
• Microsoft Graph subscriptions

​

SSL/TLS

​

Requirements

• Valid SSL certificate
• TLS 1.2 or higher
• HTTPS for all external endpoints

​

Options

• Let’s Encrypt (free)
• Commercial certificate
• Cloud provider managed certificate
• Kubernetes cert-manager

​

Domain & DNS

​

Domain Requirements

• Production domain (e.g., hetairos.yourcompany.com)
• Optional: Separate domains per environment

​

DNS Records

A record: hetairos.yourcompany.com → Load Balancer IP
CNAME: www.hetairos.yourcompany.com → hetairos.yourcompany.com

​

Storage Requirements

​

Application Storage

• Docker images: 2-5 GB
• Logs: 10-50 GB (with rotation)
• Temporary files: 5-10 GB

​

Database Storage

• User data: ~10 MB per active user
• Conversation history: ~50 MB per active user
• Growth rate: ~5-10 GB/month (100 users)

​

Monitoring & Logging

​

Metrics Collection

• CPU and memory usage
• Request latency
• Error rates
• Queue depth
• Database performance

​

Log Aggregation

• Application logs
• Error logs
• Access logs
• Audit logs

​

Tools

• Prometheus: Metrics
• Grafana: Dashboards
• ELK Stack: Logs (Elasticsearch, Logstash, Kibana)
• Azure Monitor: All-in-one (if using Azure)

​

Backup & Disaster Recovery

​

Database Backups

• Frequency: Daily (minimum)
• Retention: 30 days (minimum)
• Point-in-time recovery: Enabled
• Cross-region replication: Recommended

​

Configuration Backups

• Environment variables
• Kubernetes manifests
• SSL certificates
• Integration credentials

​

RTO/RPO Targets

• RTO (Recovery Time Objective): < 4 hours
• RPO (Recovery Point Objective): < 1 hour

​

Security Requirements

​

Authentication & Authorization

• Service account authentication
• Role-based access control (RBAC)
• API key rotation
• Secret management

​

Data Protection

• Encryption at rest (database, storage)
• Encryption in transit (TLS)
• Secrets in Key Vault/Secrets Manager
• No credentials in code or config files

​

Compliance

• GDPR (EU users)
• CCPA (California users)
• SOC 2 (enterprise)
• HIPAA (healthcare)

​

Scalability Considerations

​

Horizontal Scaling

• Stateless workers
• Load balancing
• Auto-scaling based on:
  • CPU usage
  • Memory usage
  • Queue depth

​

Vertical Scaling

• Increase pod resources
• Database tier upgrades
• Message queue tier upgrades

​

Performance Targets

• Message processing: < 5 seconds (simple)
• API response time: < 500ms (p95)
• Concurrent users: 100+ per pod
• Queue throughput: 100+ messages/sec

​

Cost Estimation

​

Small Deployment (10 users)

• Compute: $50-100/month
• Database: $25-50/month
• Queue: $10-25/month
• LLM API: $50-200/month
• Total: ~$150-400/month

​

Medium Deployment (50 users)

• Compute: $200-400/month
• Database: $100-200/month
• Queue: $25-50/month
• LLM API: $250-1000/month
• Total: ~$600-1700/month

​

Large Deployment (100+ users)

• Compute: $500-1000/month
• Database: $300-600/month
• Queue: $50-100/month
• LLM API: $500-2500/month
• Total: ~$1400-4200/month

​

Pre-Deployment Checklist

• Cloud infrastructure provisioned
• Database created and configured
• Message queue set up
• Secrets stored in Key Vault
• Domain and SSL configured
• Monitoring and logging enabled
• Backup strategy implemented
• Load balancer configured
• Network security rules set
• API keys obtained for all integrations
• Deployment manifests prepared
• CI/CD pipeline configured (optional)

​

Next Steps