0037: n8n Self-hosted vs Cloud Hosted

Status

Deferred - Reviewed and Approved, though the actual option chosen won't necessarily be implemented until we merge in the ADR for Security Guidelines for Data and AI.

Context

Our organization has been increasingly adopting n8n for workflow automation and integration tasks. As our usage grows, we should make a strategic decision about whether to continue using n8n's cloud-hosted service or migrate to a self-hosted solution on our own infrastructure.

Current Situation

We currently use n8n through their cloud platform
Usage has been growing steadily across multiple teams
We have concerns about data privacy, cost scaling, and vendor lock-in
Our AWS infrastructure is well-established and capable of hosting n8n

Decision Drivers

Cost Optimization: Evaluate long-term cost implications as usage scales
Data Privacy & Compliance: Need to ensure sensitive data remains within our control
Performance: Reduce latency for internal integrations
Customization: Ability to modify and extend n8n functionality
Infrastructure Control: Leverage existing AWS expertise and infrastructure

Considered Options

Option 1: Continue with n8n Cloud Hosted

Description: Maintain current cloud-hosted n8n service

Pros:
Zero infrastructure management
Automatic updates and maintenance
Built-in monitoring and reliability
Quick setup and deployment
Cons:
Monthly subscription costs that scale with usage
Data processed through third-party infrastructure
Limited customization options
Dependency on external service availability

Option 2: Self-host on AWS using Kubernetes

Description: Deploy n8n using Kubernetes on AWS EKS

Pros:
Full control over infrastructure and data
Scalable and resilient architecture
Integration with existing Kubernetes ecosystem
Cost-effective for high usage scenarios
Cons:
Requires Kubernetes expertise
More complex initial setup
Ongoing maintenance responsibility

Option 3: Self-host on AWS using EC2

Description: Deploy n8n on AWS EC2 instances

Pros:
Simpler deployment model
Direct control over compute resources
Familiar infrastructure management
Cons:
Manual scaling and failover management
Higher operational overhead
Less resilient than containerized solutions

Option 4: Self-host on AWS using Docker Compose

Description: Deploy n8n using Docker Compose on EC2

Pros:
Simplified container orchestration
Easy local development parity
Quick deployment and updates
Cons:
Limited to single-node deployment
Manual scaling challenges
Not suitable for high availability requirements

Decision

Host on AWS using Kubernetes

We will migrate from n8n's cloud-hosted service to a self-hosted solution using Kubernetes on AWS EKS.

Consequences

Positive Consequences

Data Privacy and Security

Complete data sovereignty: All workflow data, credentials, and execution logs remain within our infrastructure
Compliance advantages: Ideal for industries with strict compliance requirements (HIPAA, GDPR, SOC 2)
Reduced third-party risk: Eliminates dependency on n8n's security practices and data handling policies
Audit trail control: Full visibility into data access and processing

Cost Optimization

Eliminated subscription fees: No monthly costs based on workflow count or execution volume
Predictable infrastructure costs: AWS costs are more predictable and can be optimized
Economies of scale: Cost per execution decreases as usage increases
Resource optimization: Ability to right-size infrastructure based on actual needs

Performance and Reliability

Reduced latency: Direct access to internal databases, APIs, and services
Network optimization: Eliminates external network dependencies
Custom monitoring: Integration with existing monitoring and alerting systems
High availability: Kubernetes provides built-in redundancy and failover capabilities

Customization and Extensibility

Code modifications: Ability to modify n8n source code for custom requirements
Custom integrations: Direct integration with existing internal services
Version control: Ability to stay on specific versions or apply custom patches
Plugin development: Freedom to develop and deploy custom nodes

Operational Benefits

Infrastructure consistency: Leverages existing Kubernetes expertise and tooling
CI/CD integration: Seamless integration with existing deployment pipelines
Backup and recovery: Integration with existing backup strategies
Resource management: Better control over compute and storage allocation

Negative Consequences

Operational Overhead

Maintenance responsibility: We must handle all updates, security patches, and bug fixes
Infrastructure management: Ongoing monitoring, scaling, and troubleshooting
Expertise requirements: Need for Kubernetes and n8n operational knowledge
Time investment: Additional DevOps time for setup and ongoing maintenance

Complexity and Risk

Initial setup complexity: More complex deployment and configuration process
Learning curve: Team needs to understand n8n's internal architecture
Troubleshooting challenges: More complex debugging without vendor support
Integration testing: Need to test all custom modifications and integrations

Resource Requirements

Infrastructure costs: AWS EKS, compute, storage, and networking costs
Human resources: Dedicated time for maintenance and support
Monitoring tools: Additional monitoring and alerting infrastructure
Backup systems: Comprehensive backup and disaster recovery planning

Vendor Support Limitations

No vendor support: Cannot rely on n8n's support team for issues
Community reliance: Dependence on community forums and documentation
Update delays: May lag behind latest n8n releases for stability
Feature availability: May miss out on cloud-only features

Implementation Plan

Phase 1: Infrastructure Setup

Set up Kubernetes cluster on AWS EKS
Configure persistent storage and networking
Deploy n8n with basic configuration
Set up monitoring and logging

Phase 2: Migration Planning

Audit current workflows and integrations
Plan data migration strategy
Test connectivity with internal services
Develop rollback procedures

Phase 3: Pilot Deployment

Deploy n8n in staging environment
Migrate subset of workflows for testing
Validate performance and functionality
Gather feedback from users

Phase 4: Production Migration

Execute full production migration
Monitor system performance and stability
Provide user training and documentation
Establish ongoing maintenance procedures

Notes

Technical Considerations

n8n requires PostgreSQL database for production use
Kubernetes deployment should include proper resource limits and requests
Consider using AWS RDS for managed PostgreSQL
Implement proper secrets management for API keys and credentials

Security Considerations

Implement network policies to restrict access
Use AWS IAM roles for service authentication
Encrypt data at rest and in transit
Regular security updates and vulnerability scanning

Monitoring and Alerting

Set up comprehensive logging with ELK stack or similar
Monitor resource usage, performance metrics, and error rates
Implement alerting for critical failures
Regular backup testing and disaster recovery drills

References

Original Author

Dakota Washok

Approval Date

2025-10-13

Approved By

Team during engineering lead meeting

Pending

Appendix

Cost Comparison Analysis

Component	Cloud Hosted (Monthly)	Self-Hosted (Monthly)
n8n Subscription	$20-200+	$0
AWS EKS	-	$73.00
EC2 Instances	-	$150-300
RDS PostgreSQL	-	$50-100
Load Balancer	-	$20-30
Storage	-	$10-20
Total	$20-200+	$303-573

Note: Enterprise costs would be much more than $20-200 Note: Self-hosted costs are estimates based on typical usage patterns. Actual costs may vary.

Risk Assessment Matrix

Risk	Probability	Impact	Mitigation
Infrastructure failure	Medium	High	Multi-AZ deployment, monitoring
Data loss	Low	High	Regular backups, testing
Security breach	Low	High	Security scanning, access controls
Performance issues	Medium	Medium	Resource monitoring, scaling
Maintenance overhead	High	Medium	Automation, documentation