Why I Still Recommend OVH and Contabo Alongside AWS

A new client project starts, and before any actual infrastructure decisions get made, AWS is already completely assumed. Sometimes this is because a previous developer set it up that way. Sometimes it is simply because it is the only name everyone in the room knows.

Then I look at what is actually running: a Node app, a Postgres database, Redis for sessions, and Nginx in front. There is absolutely nothing that needs auto-scaling groups. There is nothing that genuinely needs twelve regions. There is nothing that is actually using more than a handful of AWS's hundreds of managed services. And the monthly bill for that handful of services is several times what the exact same compute resources would cost as plain VPS instances.

What AWS is actually good at

This is not an "AWS is bad" take. It isn't. If you genuinely need to scale from zero to a sudden, massive traffic spike and back down automatically, it is unmatched. If you are deep into managed services like RDS Aurora, Lambda, or SQS, and the tight integration between them is doing real, heavy lifting for you, it is fantastic. Or, if a client's compliance requirements specifically call for AWS, it is absolutely the right tool and completely worth the premium.

Auto Scaling Groups, multi-AZ RDS failover, and the sheer breadth of managed services solve very real problems that would otherwise need to be built, patched, and actively maintained by hand.

The brutal catch is that those features have a heavy cost whether you actively use them or not. AWS's pricing model is heavily built around granular, metered billing across dozens of different dimensions. A small EC2 instance with EBS storage, a data transfer allowance, and a small RDS instance adds up to noticeably more than the equivalent raw specs from a traditional VPS provider. It is often a factor of two to four times higher.

What a plain VPS gets you

Providers like OVH, Contabo, Hetzner, and similar sell you a strictly fixed amount of CPU, RAM, and disk space for a flat monthly price. Full stop.

There are zero data transfer surprises. There is no per-hour billing for a dozen different microscopic resource types. There is absolutely no bill that requires a spreadsheet and an AWS certification to fully understand. For a workload that is just a handful of services running fairly steadily, that flat price is both dramatically cheaper and, often much more importantly for a small business, entirely predictable.

"The server costs €40 a month" is a concrete sentence a client can actually plan their budget around. "It depends on traffic, storage IOPS, API requests, and which AZ the traffic crosses" is absolutely not.

The explicit tradeoff is that you are actively managing more yourself. There is no managed database with automatic, seamless failover. There is zero auto-scaling. There is no managed load balancer. You are manually setting those up if you actually need them.

For a massive percentage of the projects I work on, that tradeoff is completely fine. The workload simply does not need automatic failover because it is not operating at a scale where a brief downtime risk justifies the massive added complexity and cost. And the sheer time saved by not constantly configuring AWS-specific tooling more than makes up for the initial manual server setup.

How I actually decide

The question is almost never "AWS or not AWS" as some kind of absolute philosophy. The actual question is: does this specific workload genuinely need something AWS provides that a plain VPS simply doesn't, and is that specific something completely worth its high cost?

If the answer involves "we might need to scale rapidly" as a vague hypothetical rather than a strictly documented traffic pattern, a VPS that can easily be resized offline in five minutes heavily covers that risk without paying the ongoing AWS premium.

If the answer is "we are already so deep in RDS and Lambda that it would cost significantly more in engineering time to leave than it costs to stay," then that is a perfectly valid, real-world reason to stay.

For brand-new projects that do not have that gravitational pull yet, I strictly default to a VPS. I only revisit the decision if and when the workload's actual shape—not its hypothetical future shape five years from now—loudly calls for something AWS demonstrably does better. Plenty of highly profitable, incredibly reliable services run entirely on a couple of well-configured VPS instances, and the client never, ever has to explain a terrifying surprise invoice to anyone.