Offrampcloud exit calculator

Worked example · AWS

A small AWS account: EC2, RDS, S3 and the usual trimmings.

Eleven services on a three-thousand-dollar bill, with reserved instances, a Multi-AZ database and a Marketplace firewall. The kind of account where engineer time decides the answer. Read through the way the calculator reads it: the bill, what it becomes on Hetzner, the verdict, and what would change it.

The bill

$3,011 a month across 11 services after setting aside support plan, marketplace, tax, credit. The six largest lines:

ServiceGradeTodayOn HetznerChange
EC2Easy$1,435$317saves $1,118
S3Easy$476$91saves $385
RDSModerate$461$116saves $344
CloudFrontModerate$255~$38saves $217
CloudWatchModerate$150~$45saves $105
DynamoDBHard$90~$45saves $45

~ marks a rule-of-thumb ratio; the rest is priced from usage in the file.

What it becomes on Hetzner

  • 11 servers: 1× AX42-1, 4× CX33, 3× CX23, 1× CX33, 2× CX43
  • $704/mo for servers, storage and traffic
  • 56 h/mo of engineer time, $4,163
  • $26,520 one-off, no payback at this net

Extra cost per year

$22,271

Not worth moving at this size. Hetzner costs more once someone is paid 56 hours a month to run it.

Grow first: at this size the hours decide, not the hardware. 4 of your 11 services move easily; 1 is hard. 74% of the Hetzner figure is priced from your usage; the rest is ratio-priced services and engineer time, which is where the range comes from.

Likely range $6,188 more to $377 less a month. Hetzner list prices of 17 Oct 2026, default assumptions.

The same file on nine destinations

Net per year after engineer time, best first. No destination pays for this account; Hetzner loses least.

Hetzner−$22,271 over
OVHcloud−$28,440 over
Scaleway−$30,081 over
Hetzner (US)−$30,963 over
Vultr−$31,776 over
Akamai (Linode)−$32,262 over
DigitalOcean−$32,869 over
UpCloud−$35,633 over
Exoscale−$40,265 over

What would change the answer

  • Engineer time. 56 hours a month at $75/h is $4,163, 86% of the destination figure. Halve the rate or fold the work into an existing team and the verdict moves first.
  • Scale. Engineer time grows slowly with the number of servers; hardware grows with the bill. The same mix at three times the size usually flips a marginal answer.
  • Moving only part of it. The first wave (foundation: storage, traffic, edge) is S3, Elastic Load Balancing, Route 53: $405/mo for about 1 engineer days, with the databases left where they are.
  • The hard services. DynamoDB has no equivalent at the destination; keeping them on AWS removes their engineer time and their risk from the plan.

Every one of these is a setting in the report. Open the example and change them.