Tech Stack Landscape

Cloud providers offer hundreds of services across compute, storage, databases, AI, networking, and more. The naming is inconsistent across providers, which makes it hard to understand what you're looking at. This is a side-by-side map of the major categories, so you can recognize equivalent services and understand what each piece does. If Cartara flagged a cloud service in your diff, use this to understand what it is and whether there's a simpler managed alternative.

The big three cloud providers

For most builders, the choice between raw cloud providers matters less than the choice between using them directly versus using a managed platform (Vercel, Supabase, Railway) that runs on top of them. See Cloud Provider Overview and Choosing a Tech Stack for guidance on when to use which.

Compute

Running code — virtual machines, containers, and serverless functions.

What this means for builders: Lambda ≈ Azure Functions ≈ Cloud Functions (event-driven, pay per invocation). ECS/Fargate ≈ Container Apps ≈ Cloud Run (run containers without managing servers). For most apps, Vercel or Railway is simpler than any of these.

Storage

Where data lives at rest.

S3 is the de facto standard — most tools treat it as the reference implementation. Azure Blob and GCS are compatible via adapters. Cloudflare R2 is an S3-compatible alternative with no egress fees, making it popular for cost-conscious builders.

Databases

Cross-cloud alternatives that work everywhere: Supabase (PostgreSQL), PlanetScale (MySQL), Neon (serverless Postgres), MongoDB Atlas. These are often a better starting point than the raw cloud databases.

AI and machine learning

Model access: AWS Bedrock hosts Claude, Llama, and others. Azure OpenAI is the primary commercial access point for GPT-4 models with enterprise data controls. GCP Vertex AI hosts Gemini and partners.

Networking

Cloudflare sits outside this table — it's a global network layer you can put in front of any cloud provider. It's a major player in CDN, DDoS protection, WAF, and edge compute.

Security and identity

DevOps and developer tools

GitHub (owned by Microsoft) is the dominant source control and CI platform used across all clouds.

Data and analytics

Cloud-agnostic alternatives widely used across all three: Snowflake, Databricks, dbt, Fivetran.

Messaging and integration

What You'll See in Your Code

Cartara may flag use of a raw AWS service where a simpler managed alternative exists:

```javascript
// Using raw AWS S3 directly
import { S3Client, PutObjectCommand } from '@aws-sdk/client-s3'

const s3 = new S3Client({ region: 'us-east-1' })
await s3.send(new PutObjectCommand({
Bucket: 'my-bucket',
Key: uploads/${filename},
Body: fileBuffer,
}))

// Simpler if you're already on Supabase:
const { data, error } = await supabase.storage
.from('uploads')
.upload(filename, fileBuffer)
```

Both are valid — the question is whether the added control of direct S3 is worth the added complexity for your stage.

Related Concepts

• Cloud Provider Overview
• Choosing a Tech Stack
• Serverless Compute Guide
• DevOps