Cloud Vendor Product Matrix#

Choosing between cloud vendors requires mapping equivalent services across providers. AWS has 200+ services. Azure has 200+. GCP has 100+. Cloudflare has 20+ but they are tightly integrated and edge-native. This article maps the services that matter for most applications – compute, serverless, databases, storage, networking, and observability – across all four vendors with pricing, availability, and portability for each.

How to Use This Matrix#

Each section maps equivalent products across vendors, then provides:

• Pricing model: How you pay (per-request, per-hour, per-GB)
• Free tier: What you get for $0
• Portability: How easy it is to migrate to another vendor (High/Medium/Low)
• Best for: When to choose each vendor’s offering

Portability scores mean:

• High: Standard API or open protocol. Migration requires config changes only (e.g., S3 API, SQL, SMTP)
• Medium: Custom API but portable concepts. Migration requires code changes but not architecture changes (e.g., Lambda to Workers)
• Low: Deeply proprietary. Migration requires rearchitecting (e.g., DynamoDB, Cosmos DB, Durable Objects)

Global Presence#

Cloudflare is fundamentally different: code deploys to every edge location automatically. AWS, Azure, and GCP require you to choose regions and manage replication. This means Cloudflare has the lowest latency for globally distributed users by default, but you cannot pin workloads to specific regions (except with Durable Objects location hints or D1 data localization).

Serverless Compute#

Best for:

• Cloudflare Workers: Latency-sensitive APIs, edge logic, globally distributed apps. Sub-5ms cold starts are unmatched.
• AWS Lambda: Event-driven backends, complex AWS integrations (SQS, S3 triggers, Step Functions). Broadest trigger ecosystem.
• Azure Functions: .NET workloads, Microsoft ecosystem integration, Event Grid / Service Bus triggers.
• GCP Cloud Run: Docker-based workloads that need serverless scaling. Most portable (any container runs).

Container / VM Compute#

Cloudflare does not offer traditional VMs or managed Kubernetes. Cloudflare Containers (beta) run Docker workloads at the edge but are not GA and lack the isolation guarantees needed for security-sensitive workloads. If you need VMs, Kubernetes, GPUs, or bare metal, you need AWS, Azure, or GCP.

Databases#

Relational#

Best for:

• D1: Small-to-medium apps, edge-read-heavy workloads, projects where zero management matters. Not for write-heavy or large databases.
• RDS/Aurora: Production workloads needing PostgreSQL/MySQL compatibility, high write throughput, mature tooling.
• Cloud SQL: GCP-native apps, same strengths as RDS in the GCP ecosystem.

Key-Value / Document#

All key-value / document databases are low portability. Each has a proprietary API. Choosing one is a long-term commitment. KV is the simplest (just get/put with TTL) but also the most limited.

Object Storage#

Best for:

• R2: Any workload where downloads are frequent. Zero egress makes it the cheapest option at scale for serving content.
• S3: The default choice when you need the broadest ecosystem (event notifications, analytics, Glacier archiving).
• Azure Blob: Azure-native apps, Archive tier for cold storage.
• GCS: GCP-native apps, BigQuery integration for analytics on stored data.

CDN and Edge#

Cloudflare is the clear value leader for CDN. The free plan includes unlimited bandwidth, DDoS protection, basic WAF, and DNS. Every other vendor charges for bandwidth and treats DDoS/WAF as add-on services.

DNS#

DNS is fully portable – it is an open standard. Cloudflare DNS is free with fast propagation, making it the default recommendation unless you need tight integration with Route 53 health checks or Traffic Manager.

Message Queues#

Best for:

• Cloudflare Queues: Connecting Workers to Workers. Simplest setup if you are already on Cloudflare.
• SQS: The default queue for AWS workloads. Battle-tested, simple, pairs with Lambda triggers.
• Service Bus: Enterprise messaging with AMQP, sessions, and complex routing. .NET ecosystem.
• Pub/Sub: Fan-out messaging (one message, many subscribers). Analytics pipelines, event streaming.

Observability#

Cloudflare’s observability is the weakest of the four. There is no built-in tracing, limited alerting, and no equivalent to CloudWatch Dashboards. For production observability, pair Cloudflare with external tools (Datadog, Grafana Cloud, Axiom). AWS, Azure, and GCP all have comprehensive built-in observability stacks.

Identity and Access Management#

IAM is always low portability – every vendor’s permission model is deeply proprietary. Cloudflare’s IAM is simpler (API tokens scoped to services) but less granular than AWS/Azure/GCP.

Pricing Model Comparison#

The fundamental pricing difference: Cloudflare charges per operation with zero egress. AWS/Azure/GCP charge per instance-hour, per-GB-transferred, and per-operation, with egress fees on nearly everything. This makes Cloudflare extremely cost-predictable but also means you cannot optimize costs through reserved instances or committed use discounts (because there are no instances to reserve).

Free Tier Summary#

Cloudflare has the most generous permanent free tier. AWS has the broadest free tier but most of it expires after 12 months (RDS, EC2, S3, CloudFront). GCP’s free tier is mixed – some services are always-free (e2-micro, Firestore), others are trial credits. Azure is similar to AWS with 12-month limits on key services.

Portability Summary#

The most portable stack: Docker containers (Cloud Run / Fly.io) + PostgreSQL (any managed provider) + S3-compatible storage (R2 or S3) + standard DNS. This combination lets you move between vendors by changing configuration, not code.

The least portable stack: Lambda with DynamoDB triggers + SQS + CloudFront Functions + Cognito. Every component is AWS-specific. Migration requires a full rewrite.

Choosing a Vendor#

For most new projects, evaluate these criteria in order:

1. Do you need specific managed services? If your project depends on SageMaker, BigQuery, Active Directory integration, or another vendor-specific service, that vendor wins by default.

2. What is your budget? At $0-50/month, Cloudflare’s free tier and per-operation pricing is hard to beat. At $500+/month, AWS/Azure/GCP commitment discounts become significant.

3. Where are your users? If globally distributed, Cloudflare’s edge-first model gives the lowest latency without configuration. If concentrated in one region, a regional deployment on any vendor works.

4. What is your team’s expertise? The vendor your team already knows is often the right choice. Cloud expertise takes months to build. Switching vendors means retraining, not just reconfiguring.

5. How much operational overhead can you absorb? Cloudflare requires the least ops work (no regions, no VPCs, no security groups). AWS offers the most control but demands the most operational investment.

When none of these criteria produce a clear winner, default to Cloudflare for APIs and static sites (lowest cost, least ops), AWS for complex backends (broadest services), or GCP for data/ML workloads (BigQuery, Vertex AI).