2.7 KiB

Raw Blame History

Deployment Strategies for JavaScript Applications

Quick reference for deployment patterns and platforms.

Deployment Patterns

Rolling Deployment - Gradual instance replacement. Pros: Simple, zero downtime. Cons: Old/new versions mix during deploy. Best for: Most apps (default)

Blue-Green - Two identical environments, switch traffic. Pros: Instant rollback, pre-test. Cons: 2x infrastructure cost. Best for: Critical apps needing instant rollback

Canary - Route small % to new version, increase gradually. Pros: Safe real-user testing, early detection. Cons: Complex routing, needs good monitoring. Best for: High-risk changes, large user bases

Recommended Platforms

Frontend (Next.js/React)

Vercel - Zero-config Next.js, automatic HTTPS, global CDN, preview deployments, serverless functions Netlify - JAMstack focus, forms, serverless, similar to Vercel

Backend / Full-Stack

Railway - Simple, PostgreSQL/Redis/MongoDB included, GitHub integration, automatic SSL Render - Similar to Railway, good for Node.js/Docker, managed databases Fly.io - Global edge deployment, Docker-based, great for WebSocket apps

Enterprise / Complex

AWS (ECS/Fargate) - Full control, scalable, complex setup, higher cost Google Cloud Run - Serverless containers, auto-scaling, pay per use DigitalOcean App Platform - Simple Heroku alternative, reasonable pricing

CI/CD Pattern

GitHub Actions - Test on PR, deploy on merge to main GitLab CI - Similar workflow, built-in container registry CircleCI - Fast, good caching, pipeline visualization

Standard Flow:

Push to branch → Run tests
Merge to main → Build + test
Deploy to staging → Run smoke tests
Deploy to production (manual approval for critical apps)
Monitor metrics post-deploy

Environment Management

Local: .env file (never commit) Staging/Production: Platform environment variables or secrets manager Secrets: AWS Secrets Manager, HashiCorp Vault, or platform-specific

Database Migrations

Strategy: Run migrations BEFORE deploying new code Tools: Prisma migrate, TypeORM migrations, or raw SQL scripts Safety: Test migrations on staging first, backup before production

Monitoring Post-Deploy

Health check endpoints (/health, /ready)
Error tracking (Sentry, Rollbar)
Performance monitoring (Datadog, New Relic)
Log aggregation (Logtail, Better Stack)
Alert on error rate spikes or response time degradation

Rollback Plan

Immediate: Revert to previous deployment (platform UI or CLI) Database: Keep backward-compatible migrations for at least one version Feature Flags: Use feature flags to disable new features without redeployment

2.7 KiB Raw Blame History