Kubernetes vs Docker Compose for a 3-service product on a single VM
Scalability ↔ Simplicity
Context
Choose the deployment model for an early-stage product.
Three services (API, worker, web) plus Postgres and Redis. Under 50 req/s. A contractor proposed EKS "so we're ready to scale". Monthly cost would go from ~$80 to ~$400 before any workload.
Is there a middle ground (ECS, Fly.io, Nomad, k3s) or is Compose on a VM with a good backup story genuinely fine at this stage?
Constraints
- Team
- 3 engineers
- Traffic
- < 50 req/s
- Budget
- $300/month
KubernetesvsDocker Compose
Community verdict
8 engineers · 4 opinions
With these constraints, what would you choose?
One choice per engineer. You can change it any time.
Simplicity →
Trade-offs
Dimensions
- Scalability
- 5Kubernetes scores 5 of 52Docker Compose scores 2 of 5
- Simplicity
- 1Kubernetes scores 1 of 55Docker Compose scores 5 of 5
- Self-healing
- 5Kubernetes scores 5 of 5
Community discussion
4 comments
Have you made this decision in production? Share your reasoning.
Sign in to commentWe ran Compose on one VM for three years and 40 paying customers. The day we needed Kubernetes, we knew exactly why, and the migration took two weeks because the services were already containerised.
If anyone on the team already knows Kubernetes, a managed control plane is less scary than it looks — and you get rolling deploys and health checks for free. But EKS at $400/month for 50 rps is hard to justify.
ECS Fargate is the honest middle ground on AWS. Rolling deploys and health checks without running a cluster.
Compose + systemd + automated backups + a second VM for blue/green will carry you to real product-market fit. Re-evaluate when you need more than one machine for the workload, not for the resume.