I got everything off the cloud and am paying less
An indie developer claims a 10× infrastructure cost reduction by moving a small web stack (web app, Postgres, Redis) from AWS to rented dedicated servers, reigniting debate over when cloud services are worth their premium. Commenters weigh direct savings against hidden labor, downtime risk, and “Day 2” concerns like backups and disaster recovery, while also highlighting issues of vendor lock‑in, loss of operational know‑how, and the value of retaining control over one’s infrastructure.
Cost Savings vs Time and Labor
- Many agree the raw dollar savings (~$15k/year, “10x less”) is meaningful for bootstrapped or indie projects but small relative to even one engineer’s salary.
- Critics stress “hidden” costs: time spent designing, migrating, maintaining, and handling incidents; ongoing “day-2” tasks like backups, DR, scaling, monitoring.
- Counterpoint: several claim real-world maintenance for a few bare-metal/VPS servers is hours per year, not per week, especially for simple stacks.
Risk, Reliability, and SLAs
- Pro-cloud view: hyperscalers absorb hardware failures, networking, power, and 24/7 coverage that would be expensive to replicate; outages at AWS/Azure are widely understood by customers.
- Skeptics: cloud doesn’t eliminate outages; many report fewer incidents on their own boxes than via cloud dependencies, and say on-prem/VPS can be designed with redundancy, UPS, etc.
- Some stress that outage cost is often overestimated; a brief, well-communicated downtime rarely destroys a business.
Cloud Complexity vs Bare-Metal Complexity
- Several note AWS pricing and service sprawl are confusing enough that you need specialists; “accidentally paying 10x” is a real risk.
- Others argue that bare metal / VPS is at least as complex once you consider backups, failover, storage, and scaling.
- There is debate over whether cloud actually reduces ops headcount or just shifts it to DevOps/SRE with different tools.
Scale and Use Cases
- Consensus: for very small deployments, big-cloud is often overkill; a single VPS or rented dedicated server can be simpler and cheaper.
- At large scale or with strict SLAs/DR requirements, cloud and managed services can still make economic and operational sense.
- Many emphasize the middle ground: rented bare-metal or VPS in data centers (Hetzner, OVH, colo) vs building your own data center.
Skills, Control, and “Propaganda”
- Some see big-cloud as “deskilling” engineers and creating lock-in; they value Linux/sysadmin skills and control over hardware and network costs.
- Others explicitly choose cloud because they don’t want to care about servers; their time is more valuable than the savings.
- A few suggest strong industry marketing and incentives bias discourse toward cloud, while “exit the cloud” stories push back against that narrative.