Snapshot or backup: recovery is what matters
Check what a copy protects, how independent it is and whether you can actually restore the service.
Compare maintenance, backups, recovery and total cost before choosing a VPS or a managed hosting service.
THE USEFUL ANSWER
Assess a VPS against your ability to administer it, not just its resources. For managed hosting, check exactly what the supplier covers. The best fit has clear responsibilities that you can sustain.
A hosting price does not describe the work you still need to do. List the operating system, application, certificates, patches, backups, monitoring and support. Put an owner against each line.
AWS’s shared-responsibility model illustrates how the split depends on the service: outsourcing infrastructure does not automatically transfer every responsibility. It helps frame the questions, but the terms of your chosen supplier remain the reference.
Ask both suppliers these questions. A vague answer is a reason to clarify the scope, not a reason to assume coverage.
| Topic | Practical question |
|---|---|
| Updates | Who patches the operating system and the application? |
| Backups | Which data, which retention and which access? |
| Recovery | Who restores the service, how and on what stated timeline? |
| Support | Which incidents are covered and during which hours? |
| Exit | How can files and data be exported to another supplier? |
For an initial budget, add infrastructure, backups, required options and administration time. Keep the distinction clear: the value of your time is not a bill from the supplier, but it matters when planning the project.
Use figures from a dated offer or quotation. Keep the tax basis consistent across all lines. Our hosting cost calculator separates cash expenses from valued time; it does not fetch market prices or promise a particular bill.
Consider an editorial site with articles, images, local search and merchant links, but no user accounts. Our first scenario would be to generate the pages ahead of time and host the resulting files. A persistent application stack is not automatically required for that particular need.
Accounts, personalised information or server-side processing change the discussion. Specify those features before buying computing power. They can justify extra services; they do not automatically justify the most complicated offer.
Who applies operating-system updates? Who patches the application? Who checks alerts? Who restores the service after a failed update? Who steps in when you are unavailable? We would ask for written answers before committing.
“Managed” does not replace this list. Examine contractual scope, exclusions and support conditions. Having administrator access is also not proof that you want to take on all the duties attached to it.
Write a one-page migration procedure: where files and copies are held, how authorised people retrieve access and how the result will be checked. A project dependent on one person should at least have documentation another person can use.
Before moving an important site, practise restoration to a separate destination. Check that required files, data and configuration can be recovered without overwriting production. Our backup and snapshot guide expands on that exercise. Clear ownership and a believable recovery procedure are more useful than two columns of memory and processor specifications.