ADR-003: Multi-Cloud Strategy

Status

Accepted

Context

Implementation Studio initially launched with GCP-only support for cloud infrastructure. As the platform evolved, we faced a strategic decision:

Market reality - AWS is the largest cloud provider (32% market share), many customers use AWS
Completeness - Multi-cloud support would make the platform more valuable and applicable
Learning value - Teaching cloud-agnostic thinking and highlighting provider differences
Competitive advantage - Most tutorials focus on one cloud; multi-cloud expertise is valuable
Maintenance burden - Adding AWS would double modules, documentation, and testing
Scope question - If we add AWS, should we also add Azure? Where does it end?

The decision impacts:

Module development - Need AWS equivalents for GCP modules
Lab updates - Labs must support multiple providers
Documentation - Provider comparisons, migration guides, feature parity
Maintenance - Ongoing maintenance of multiple provider implementations
Learning experience - Learners can choose their preferred cloud provider

Decision

We will add AWS support to Implementation Studio, creating a GCP + AWS multi-cloud platform. We will not add Azure support at this time.

AWS Support Scope

Core Infrastructure Modules (Phase 1):

✅ EKS Cluster (equivalent to GKE)
✅ VPC (equivalent to GCP VPC)
✅ VPC Private (equivalent to GCP Private VPC)
✅ ECR (equivalent to Artifact Registry)
✅ Security Groups (equivalent to Firewall Rules)
✅ RDS (equivalent to Cloud SQL)

Lab Support:

✅ Lab 01: Standard Deployment (GCP + AWS)
✅ Lab 03: Private Network Deployment (GCP + AWS)
✅ Lab 04: Firewall-Restricted Deployment (GCP + AWS)
✅ Lab 05: POC Sprint (Kind + GCP + AWS)
✅ Lab 06: Multi-Tenant Deployment (Kind + GCP + AWS)
✅ Lab 07: Integration Patterns (GCP + AWS)
✅ Labs 02, 08, 09: Cloud-agnostic (no changes needed)

Azure Decision

We will NOT add Azure support for the following reasons:

Market coverage - GCP + AWS covers ~60% of cloud market
Maintenance burden - Adding Azure would triple maintenance (GCP + AWS + Azure)
Focus - Better to excel at GCP + AWS than be mediocre at all three
Community - Azure support can be a community contribution if demand exists
Resource constraints - Limited time/resources to maintain three providers

Consequences

Positive

Broader applicability - Platform useful for AWS customers (largest market share)
Learning value - Teaches cloud-agnostic thinking and provider differences
Competitive differentiation - Most tutorials focus on one cloud; we cover two
Real-world relevance - Many customers use AWS; skills are transferable
Portfolio value - Demonstrates multi-cloud expertise (valuable for career)
Flexibility - Learners can choose their preferred cloud provider
Feature comparison - Highlights differences between providers (educational)

Negative

Maintenance burden - 2x modules, 2x documentation, 2x testing
Complexity - Different patterns (EKS vs GKE), terminology, best practices
Documentation overhead - Must document provider differences and migration paths
Feature parity - Must keep features in sync between providers
Testing complexity - Must test both providers, manage two sets of credentials
Learning curve - More documentation to navigate, potential confusion for beginners

Neutral

Market coverage - GCP + AWS covers ~60% of market (good, not perfect)
Future expansion - Azure can be added later if demand exists
Community contributions - Azure support can be community-driven

Alternatives Considered

Option 1: GCP Only (Status Quo)

Pros:

Simpler - Single provider, less complexity
Faster development - No need to maintain AWS modules
Lower maintenance - Half the modules, documentation, testing
Focused - Can excel at GCP patterns
Lower barrier - Easier for beginners (one cloud to learn)

Cons:

Limited applicability - Only useful for GCP customers
Missed opportunity - AWS is largest cloud provider (32% market share)
Less valuable - Skills less transferable
Competitive disadvantage - Other platforms may offer multi-cloud
Learning limitation - Doesn't teach cloud-agnostic thinking

Why not chosen: While simpler, GCP-only limits the platform's applicability and learning value. AWS support significantly increases the platform's value proposition.

Option 2: AWS Only

Pros:

Largest market - AWS has 32% market share
Broader applicability - More customers use AWS
Simpler - Single provider, less complexity
Focused - Can excel at AWS patterns

Cons:

Abandon GCP investment - Would lose all GCP modules and labs
Less differentiation - Many tutorials already focus on AWS
Learning limitation - Doesn't teach cloud-agnostic thinking
Waste of effort - GCP modules already built and working

Why not chosen: Abandoning GCP would waste existing investment. GCP + AWS provides better coverage and learning value.

Option 3: GCP + AWS + Azure (All Three)

Pros:

Complete coverage - Covers all major cloud providers
Maximum applicability - Useful for any cloud customer
Comprehensive - Most complete learning platform

Cons:

3x maintenance burden - Triple the modules, documentation, testing
Quality risk - Harder to maintain quality across three providers
Resource constraints - Limited time/resources to maintain three providers
Diminishing returns - GCP + AWS covers ~60% of market; Azure adds ~20%
Complexity - More documentation, more patterns, more confusion
Focus dilution - May spread effort too thin

Why not chosen: The maintenance burden of three providers outweighs the benefits. GCP + AWS covers the majority of the market, and Azure can be added later if demand exists.

Option 4: Cloud-Agnostic Abstraction Layer

Pros:

Single interface - One module interface, multiple backends
Easier expansion - Adding new providers is easier
Consistent patterns - Same patterns across all providers
Less duplication - Shared logic, provider-specific implementations

Cons:

Complexity - Abstraction layer adds complexity
Hides differences - May hide important provider-specific differences
Learning limitation - Doesn't teach provider-specific patterns
Maintenance overhead - Abstraction layer itself requires maintenance
Less educational - Learners don't see provider differences clearly

Why not chosen: While elegant, an abstraction layer would hide important provider differences and reduce educational value. Direct provider modules are clearer and more educational.

Option 5: Parallel Tracks (Separate GCP/AWS Labs)

Pros:

Clear separation - Learners follow one track (GCP or AWS)
No confusion - No need to choose provider in each lab
Focused learning - Can complete without learning both

Cons:

Duplicate content - Labs duplicated for each provider
More maintenance - Duplicate labs to maintain
Harder to sync - Difficult to keep labs in sync
Less flexible - Can't easily switch providers

Why not chosen: Duplicating labs would create significant maintenance burden. Conditional module selection (single lab, choose provider) is more maintainable.

Decision Rationale

GCP + AWS was chosen because it:

Maximizes market coverage - GCP + AWS covers ~60% of cloud market
Balances value and maintenance - Good coverage without excessive maintenance burden
Teaches cloud-agnostic thinking - Learners see provider differences and similarities
Demonstrates expertise - Multi-cloud skills are valuable and differentiate the platform
Maintains quality - Two providers is manageable; three would risk quality
Allows future expansion - Azure can be added later if demand exists

Azure exclusion was decided because:

Diminishing returns - GCP + AWS covers majority of market; Azure adds less value
Maintenance burden - Three providers would triple maintenance
Quality risk - Harder to maintain quality across three providers
Resource constraints - Limited time/resources for three providers
Community option - Azure can be community contribution if demand exists

Implementation

Phase 1: Core AWS Modules ✅ Complete

Created AWS equivalents for core GCP modules:

modules/aws/eks-cluster/ - EKS cluster (equivalent to GKE)
modules/aws/vpc/ - VPC with public/private subnets
modules/aws/vpc-private/ - Private VPC with VPC endpoints
modules/aws/ecr/ - Elastic Container Registry
modules/aws/security-groups/ - Security groups for strict egress
modules/aws/rds/ - RDS for database integration

Phase 2: Lab Updates ✅ Complete

Updated labs to support both providers:

Lab 01 - Standard deployment (GCP + AWS)
Lab 03 - Private network (GCP + AWS)
Lab 04 - Firewall-restricted (GCP + AWS)
Lab 05 - POC Sprint (Kind + GCP + AWS)
Lab 06 - Multi-tenant (Kind + GCP + AWS)
Lab 07 - Integration patterns (GCP + AWS)

Phase 3: Documentation ✅ Complete

Created comprehensive multi-cloud documentation:

docs/02-multi-cloud/provider-comparison.md - Technical GCP vs AWS comparison
docs/02-multi-cloud/migration-guide.md - Migration instructions
docs/02-multi-cloud/feature-parity-matrix.md - Feature comparison
Updated all lab documentation with provider selection guides

Module Parity Strategy

Keep modules functionally equivalent:

Same variable names where appropriate
Same output structure
Document differences clearly
Provide migration guides

Example:

hcl

# GCP
module "gke_cluster" {
  source = "../../modules/gcp/gke-cluster"
  project_id = var.project_id
  # ...
}

# AWS
module "eks_cluster" {
  source = "../../modules/aws/eks-cluster"
  # Equivalent variables
  # ...
}

Provider Selection in Labs

Labs use conditional module selection:

hcl

variable "cloud_provider" {
  description = "Cloud provider: gcp or aws"
  type        = string
  validation {
    condition     = contains(["gcp", "aws"], var.cloud_provider)
    error_message = "Cloud provider must be 'gcp' or 'aws'."
  }
}

module "cluster" {
  source = var.cloud_provider == "gcp"
    ? "../../modules/gcp/gke-cluster"
    : "../../modules/aws/eks-cluster"
  # ...
}

Future Considerations

Azure Support

Decision: Not adding Azure support at this time.

Future consideration: Azure support can be added if:

Community demand exists
Resources become available
GCP + AWS are successful and well-maintained

Approach: Azure support would be a community contribution, not core platform feature.

Other Cloud Providers

Decision: Focus on GCP + AWS only.

Rationale: GCP + AWS covers majority of market. Adding more providers (Oracle Cloud, IBM Cloud, etc.) would dilute focus without significant value.

References

Multi-Cloud Considerations - Strategic analysis
Provider Comparison Guide - Technical comparison
Migration Guide - Migration instructions
Feature Parity Matrix - Feature comparison

Date: January 5, 2026
Author: Ben Hankins
Status: Accepted

ADR-003: Multi-Cloud Strategy ​

Status ​

Context ​

Decision ​

AWS Support Scope ​

Azure Decision ​

Consequences ​

Positive ​

Negative ​

Neutral ​

Alternatives Considered ​

Option 1: GCP Only (Status Quo) ​

Option 2: AWS Only ​

Option 3: GCP + AWS + Azure (All Three) ​

Option 4: Cloud-Agnostic Abstraction Layer ​

Option 5: Parallel Tracks (Separate GCP/AWS Labs) ​

Decision Rationale ​

Implementation ​

Phase 1: Core AWS Modules ✅ Complete ​

Phase 2: Lab Updates ✅ Complete ​

Phase 3: Documentation ✅ Complete ​

Module Parity Strategy ​

Provider Selection in Labs ​

Future Considerations ​

Azure Support ​

Other Cloud Providers ​

References ​

ADR-003: Multi-Cloud Strategy

Status

Context

Decision

AWS Support Scope

Azure Decision

Consequences

Positive

Negative

Neutral

Alternatives Considered

Option 1: GCP Only (Status Quo)

Option 2: AWS Only

Option 3: GCP + AWS + Azure (All Three)

Option 4: Cloud-Agnostic Abstraction Layer

Option 5: Parallel Tracks (Separate GCP/AWS Labs)

Decision Rationale

Implementation

Phase 1: Core AWS Modules ✅ Complete

Phase 2: Lab Updates ✅ Complete

Phase 3: Documentation ✅ Complete

Module Parity Strategy

Provider Selection in Labs

Future Considerations

Azure Support

Other Cloud Providers

References