Don't Buy a House for an MVP: The Case for Renting Your Infrastructure
A Common Mistake
I see it constantly. A startup gets their first round of funding, hires an infrastructure team, and builds a distributed microservices architecture with multi-region failover, blue-green deployments, and a 99.99% SLA — for an app with 200 users.
They've bought a house when they needed to rent a room.
The Infrastructure Ladder
Good infrastructure architecture matches your current stage, not your ambitions. Here's how I think about it:
Stage 1: Prove the Concept (0–1K users)
- Single server or managed PaaS (Railway, Render, Heroku)
- Managed database (PlanetScale, Supabase, RDS)
- No load balancer, no auto-scaling, no multi-AZ
- Goal: Ship and learn, not survive a DDoS
Stage 2: Stabilise (1K–50K users)
- Move to AWS/GCP with a proper VPC
- Introduce a load balancer (ELB or ALB)
- Add auto-scaling groups — but only for your compute layer
- Set up basic monitoring and alerting
- Goal: Don't go down during your first viral moment
Stage 3: Scale (50K+ users)
- Multi-AZ deployments for critical services
- CDN for static assets and caching
- Read replicas for the database
- Begin separating services where bottlenecks appear
- Goal: Handle 10x your current load without a rewrite
Stage 4: Enterprise (mission-critical workloads)
- Multi-region active-active
- Chaos engineering
- Full IaC (Terraform, Pulumi)
- SOC 2, compliance, audit trails
- Goal: Reliability and compliance at scale
The Cost of Over-Engineering
When you over-build early, you pay in three ways:
- Money: Cloud bills that dwarf your revenue
- Speed: Every feature takes 3x longer to ship
- Complexity: Onboarding new engineers is painful, debugging is slow
What to Actually Do
Use managed services aggressively. AWS RDS, ElastiCache, SQS, S3 — these are not "beginner" tools. They're what senior engineers choose because they've learned the hard way.
Save your engineering hours for what's unique to your business. Nobody gets a competitive advantage from managing their own database replica set.
Rent now. Buy (own) later.
Ishan is an AWS-experienced cloud architect who has led infrastructure migrations for enterprise organisations. Drop him a message if you're thinking through your architecture.