Galleon
Sign In

Deployment Tiers

Galleon picks the right AWS infrastructure for your app based on what kind of workload it is. Two shapes, four sizes.

Backend services — FastAPI, Django, Express, and other containerized apps — deploy to ECS Fargate. Pick the size that fits your traffic.

Frontends — Next.js applications — deploy serverless to Lambda, CloudFront, and S3. One size, scales automatically.

Backend Tiers

For FastAPI, Django, Express, and other containerized applications.

Starter

  • Estimated AWS cost: $25--45/month
  • Infrastructure: ECS Fargate + Application Load Balancer
  • CPU / Memory: 0.25 vCPU / 512 MB
  • Scaling: 1--3 instances, always-on minimum of 1
  • Networking: Public subnets only (no NAT gateway)
  • Best for: MVPs, side projects, low-traffic apps

The simplest container deployment. A minimal ECS Fargate service behind an Application Load Balancer with health checks and auto-restart. Runs in public subnets to avoid NAT gateway costs, keeping the bill low while still giving you a production-grade container runtime with CloudWatch logging.

Growth

  • Estimated AWS cost: $80--180/month
  • Infrastructure: ECS Fargate + Application Load Balancer
  • CPU / Memory: 0.5 vCPU / 1 GB
  • Scaling: 1--5 instances, always-on minimum of 1
  • Best for: Production apps with sustained traffic
  • Optional add-ons: RDS PostgreSQL, ElastiCache Redis

The default for production workloads. Always-on minimum keeps response times consistent — no cold starts, no scale-from-zero latency. Auto-scales up to handle traffic spikes.

Scale

  • Estimated AWS cost: $200--500/month
  • Infrastructure: ECS Fargate + Application Load Balancer, multi-AZ
  • CPU / Memory: 1 vCPU / 2 GB
  • Scaling: 2--10 instances, always-on minimum of 2
  • Best for: High-traffic, mission-critical applications
  • Optional add-ons: RDS PostgreSQL, ElastiCache Redis, CloudFront CDN

Built for workloads with SLA requirements. Multi-AZ deployment for fault tolerance, larger compute footprint, and a higher minimum instance count for redundancy.

Backend tier comparison

StarterGrowthScale
AWS cost$25--45/mo$80--180/mo$200--500/mo
ComputeECS Fargate (public subnets)ECS FargateECS Fargate (multi-AZ)
CPU / Memory0.25 vCPU / 512 MB0.5 vCPU / 1 GB1 vCPU / 2 GB
Min instances112
Max instances3510
Database (RDS)--OptionalOptional
Cache (Redis)--OptionalOptional
CDN----Optional

Frontend Tier

For Next.js applications.

Serverless

  • Estimated AWS cost: $1--20/month
  • Infrastructure: Lambda + CloudFront + S3
  • Scaling: Automatic, zero to high traffic
  • Available for: Next.js

Galleon uses OpenNext to compile your Next.js app for AWS Lambda. The deployment includes four Lambda functions — server, image optimizer, ISR revalidation, and warmer — plus a CloudFront distribution and S3 bucket for static assets.

CloudFront caches aggressively at the edge, so most read traffic is served without invoking Lambda. The result is a deployment that costs almost nothing at low traffic and scales smoothly to high traffic without configuration.

How Galleon picks a tier

Galleon recommends a tier when you connect your repository, based on what it detects in your code.

Is your app a Next.js frontend? → Serverless.

Otherwise (FastAPI, Django, Express, containerized React, etc.), Galleon picks a backend tier based on production signals:

  • Starter — default for simple applications with no production indicators
  • Growth — database, authentication, structured logging, or background jobs detected (2+ indicators)
  • Scale — cache layers, worker processes, large data volumes, or 4+ production indicators

The recommendation is a starting point. You can pick any tier from the deployment configuration — Galleon will provision whatever you choose.

Changing tiers

You can change your application's tier at any time from the repository detail page. Select a new tier from the deployment configuration and click Deploy.

Between backend tiers (Starter ↔ Growth ↔ Scale): Galleon updates the Terraform configuration and triggers a redeploy on the new infrastructure. Brief downtime during the cutover; your old infrastructure is destroyed once the new deployment passes health checks.

Next Steps