If you are writing a backend, a mobile application, or an MVP, then knowledge of the cloud will save you weeks of setup, money on infrastructure, and stress when scaling. Below is a condensed map of the area, without magic and with practice.
Why is the cloud about speed and not just DevOps?
Quick start: minutes instead of days — raised the DB, API and queue and roll in prod.
Scaling: autoscalers and load balancers are experiencing peak loads.
Payment upon completion: don't buy hardware, don't service racks.
Safety out of the box: IAM, KMS, private networks — fewer homemade crutches.
Bottom line: clouds are about releasing features faster and more stable. The sooner you understand the basics, the easier it is to go further.
Mini map of service matches
Study the Compute + Storage + Database "triad", add IAM + VPC + Monitoring — and you can already assemble a production minimum.
Task | AWS | GCP | Yandex Cloud |
|---|---|---|---|
VM (IaaS) | EC2 | Compute Engine | Compute Cloud |
Object storage | S3 | Cloud Storage | Object Storage |
Managed Kubernetes | EKS | GKE | Managed Service for Kubernetes |
Serverless functions | Lambda | Cloud Functions / Cloud Run | Cloud Functions |
Container PaaS | ECS / App Runner | Cloud Run | Serverless Containers |
Relational DB | RDS (PostgreSQL/MySQL) | Cloud SQL | Managed PostgreSQL / MySQL |
NoSQL | DynamoDB | Firestore / Bigtable | YDB |
Event queues/bus | SQS / EventBridge | Pub/Sub | Message Queue / EventHub |
Observability | CloudWatch / X-Ray | Cloud Logging / Trace | Cloud Logging / Monitoring |
Secrets and access | IAM / Secrets Manager | IAM / Secret Manager | IAM / Lockbox |
CDN | CloudFront | Cloud CDN | Cloud CDN |
What a developer needs to know at the start
VPC, Subnets, Security Groups/Firewall, NAT, Load Balancer
S3/Cloud Storage/Object Storage, managed PostgreSQL, Redis cache
Docker + Kubernetes/Cloud Run/App Runner or features
Secrets Manager / Lockbox, Least Privilege
CI/CD: build → tests → scan → deploy (container registry)
Logs, metrics and alerts (5xx errors, latency, budget)
Basic web application architecture
Frontend on static (Object Storage + CDN)
Backend in a container (Cloud Run / GKE / ECS) with auto scaling
Database: managed PostgreSQL, read replicas
Redis cache for sessions and rate limiting
Files — S3-compatible storage + presigned URLs
Queues/background processing: Pub/Sub or SQS
Observability: logs, metrics, tracing + alerts to the messenger
Serverless vs Kubernetes vs VM
Serverless — minimum DevOps, pay for calls. Ideal for APIs, bots, webhooks, and cron tasks.
Kubernetes — flexibility and control, justified with growth and several services. Requires experience.
BM — full control of the environment for specific tasks. More manual work.
Common mistakes and how to avoid them?
Public buckets by default → close access, use presigned URLs.
Secrets in environment variables → store in Secrets Manager / Lockbox, enable rotation.
No alerts → add thresholds for 5xx, response time and costs.
One database without backups → enable automatic backups, PITR, and check recovery.
Hard vendor lock-in → standardize through Docker, OpenAPI, and Terraform.
In the attachment Code We have fascinating courses — without unnecessary theory. We analyze the basics and make a mini-project. In our Telegram channel — Discussions of architectures and analysis of typical rakes.