2-Week GCP Migration Enables Multi-Cloud Distribution at Scale

How Refold AI leveraged Facets to transform their AWS-native integration agent platform into a multi-cloud distributed solution, enabling rapid deployment across AWS, GCP, and on-premise environments for their global distributor network.

Strategic Value Delivered by Facets

Before

• AWS-Native Lock-in: Originally built entirely on AWS, limiting customer acquisition to AWS-only enterprises

• Manual Distribution Model: Each distributor and customer deployment required manual infrastructure setup

• Day-2 Operations Chaos: Tracking microservice versions across multiple customer environments was becoming unmanageable

• GCP Expertise Gap: No in-house GCP knowledge despite customer demand for multi-cloud support

After

• True Multi-Cloud Platform: Seamless deployments across AWS, GCP, and on-premise from a single codebase

• Self-Service Distribution: Partners provision customer environments through an automated console

• Central Command Center: Complete visibility and control over all global deployments

• Flexible Resource Management: Dynamic environment scaling based on customer usage patterns

The Challenge: From AWS-Native to Multi-Cloud Distribution

Refold AI, an AI-native integration platform that replaces traditional system integrators with intelligent agents, faced a critical scaling challenge. Their platform—handling enterprise integrations for SAP, Oracle Fusion, Workday, and other complex systems—was built entirely on AWS. The architecture included:

• Multiple microservices for their AI agent orchestration
• Various databases supporting different integration patterns
• Cloud storage buckets for data processing
• Redis for caching and state management

As Refold expanded to serve 30+ paying enterprise customers, including Incorta and Naehas, their distribution model created unique infrastructure challenges:

1. Cloud Diversity Requirements: Enterprise customers demanded deployment options beyond AWS—many required GCP or even on-premise installations
2. Distributor Multiplication: Each distributor partner needed their own infrastructure, and each of their customers needed isolated environments
3. AWS-to-GCP Portability: Being AWS-native, they lacked GCP expertise and faced significant re-architecture challenges
4. Sandbox Explosion: Development and testing requirements multiplied the number of environments needed

How Facets Transformed Refold's Distribution Architecture

From AWS-Native to Multi-Cloud Blueprint Architecture

Facets enabled Refold to maintain its AWS-native advantages while expanding to other clouds:

AWS Blueprint (Original):

• Leveraged existing AWS-native modules
• Preserved all optimizations and configurations
• Maintained as the primary blueprint

GCP Blueprint (New Development):

• Initially explored AWS S3 connectivity for minimal changes
• Facets team developed MinIO + GCS integration for a full GCP-native experience
• Later optimized to use native GCS with S3-compatible endpoints
• Achieved feature parity with AWS deployment

On-Premise Blueprint (Innovation):

• K3s Kubernetes deployment on just two VMs
• MinIO for object storage with disk backup
• Comprehensive backup strategies for databases and storage
• Enabled deployment even in air-gapped environments

All blueprints were strategically organized:

• Same repository with different branches for each cloud
• Identical artifacts and release streams
• Cloud-specific optimizations while maintaining application consistency

Lift-and-Shift Discovery for GCP Migration

The migration from AWS to GCP was accelerated through:

• Infrastructure Discovery: Facets discovered their entire AWS setup
• Blueprint Generation: Automated creation of equivalent GCP configurations
• Progressive Migration: Initially considered cross-cloud S3 access, then moved to full GCP-native services

Central Command Center: Infrastructure as a Product

Refold built a partner console on top of Facets that transformed infrastructure into a product feature:

Partner Self-Service Portal:

• OEM partners request new customer deployments
• Automated provisioning through Facets APIs
• Real-time deployment status and monitoring

Release Management Dashboard:

• Track which microservice versions are deployed where
• Centralized control for updates across all environments
• One-click rollback capabilities

Advanced Deployment Patterns

Facets enabled sophisticated deployment models that matched Refold's diverse customer needs:

Hybrid Cloud-VM Deployments:

• Deploy Kubernetes clusters on VMs when full cloud access isn't available
• Support customers with limited cloud permissions
• The same blueprint works on cloud VMs or bare metal

Multi-Tenancy Options:

• Single Kubernetes cluster with namespace isolation
• Dedicated clusters per customer
• Flexible resource allocation based on customer tier

Key Technical Achievements

GCP Portability

• 2 weeks to GCP portability from AWS

Environment Lifecycle Management

• Scalable Active Environments: Flexible scaling and hibernation
• On-Demand Creation: Spin up test environments for demos
• Cost Optimization: Scale down or destroy unused environments
• Quick Recovery: Recreate environments from blueprints when needed

Release Velocity

• Daily Deployments: Multiple releases tracked centrally
• Version Control: Complete visibility of what's deployed where
• Rollback Safety: Quick recovery from any issues

Cloud Flexibility

As Refold supports AWS, GCP, Azure, and air-gapped on-premise environments, Facets enabled true deployment flexibility:

• Customers can choose their preferred cloud
• Gradual migration paths between clouds
• No vendor lock-in concerns

Business Impact

Accelerated Distribution

• Partners can now provision environments instantly
• Reduced customer onboarding from weeks to hours
• Enabled global expansion without operational overhead

Market Expansion

• From AWS-Only to Multi-Cloud: No longer losing deals due to cloud requirements
• On-Premise Capability: Can serve highly regulated industries
• Global Reach: Deploy in any region with any cloud provider

Operational Excellence

• Central Visibility: Single pane of glass for all deployments
• Guaranteed Consistency: Drift-free environments across all clouds
• Reduced Complexity: One team manages multiple cloud platforms

What Made the Difference?

AWS-Native Preservation: Unlike typical multi-cloud solutions that require lowest-common-denominator approaches, Facets allowed Refold to maintain AWS-native optimizations while adding GCP and on-premise capabilities.

GCP Expertise Bridge: Facets team provided the GCP knowledge Refold lacked, developing custom modules and handling cloud-specific optimizations.

Distribution Platform Integration: By building their partner console on Facets APIs, Refold turned infrastructure complexity into a competitive advantage.

Supporting Refold's Mission

Facets' approach aligns perfectly with Refold's vision of making enterprise integrations "cheap, fast, and invisible"—now achievable across any cloud or deployment model their customers require.

Future Growth Platform

With Facets as their infrastructure backbone, Refold is positioned for unlimited scale:

• Partner Ecosystem Growth: Onboard more distributors without operational overhead
• Customer Flexibility: Support any deployment preference from managed cloud to air-gapped on-premise
• Geographic Expansion: Deploy in any region with local cloud providers
• Focus on Innovation: Engineering team focuses on AI agent development rather than infrastructure management