Managed infrastructure for SMBs
How to Run a Backup Restore Test: A Small Business Checklist
The short answer
A backup only counts once you have brought something back from it. A restore test is a planned exercise where you pick a system, restore it into a safe place away from your live servers, check that it actually works, time how long it took and write the result down. Small businesses can run a useful test in an afternoon. This guide covers what to test, where to restore it, how to check the result, what to write in your record and how often to repeat it, with a checklist you can use on your next test.
1. Why a green backup job is not proof
Backup software reports that a job finished. It does not tell you that the files inside are complete, that the database will start, that the application still knows its passwords and licence keys, or how many hours a real recovery would take. Problems such as a skipped folder, an expired encryption key, a backup that only covers part of a disk or a copy that sits on the same server it protects usually stay hidden until the day you need it. The only way to find them early is to restore on purpose, when nothing is on fire.
2. Decide what to test first
List the systems your business could not run without for a day: the accounting or ERP database, the line of business application, the file share, the website or online store, the mail system. For each one, write down two numbers you can live with: how much recent data you could afford to lose, and how long you could be without it. Start your testing with the system where both numbers are smallest. One well documented test of your most important system is worth more than a quick look at ten.
3. Restore into a separate place, never over live servers
Never test by restoring on top of a production server. Use a separate virtual machine, a spare server or an isolated test network, and keep it off your normal network so a restored copy cannot send mail, sync files or talk to customers by mistake. Give the test machine its own name and address, and make sure nobody points real users at it. If your backups are encrypted, this is also where you find out whether the people who would run a real restore can find the key.
4. Check that it works, not just that it copied
A restore is successful when the system does its job, not when the files appear. Open a handful of recent documents. Start the database and run a report you know the answer to. Log in to the application and look at the newest records to see how fresh the data is. For a website, load the key pages and confirm the shop or forms work in test mode. Write down anything that needed a manual fix, such as a password you had to look up or a service that needed a setting changed, because those are the steps that slow a real recovery.
5. Time it from start to finish
Start the clock when you begin finding the backup and stop it when someone could use the system again. Include the time to locate the right copy, download or move it, rebuild the server, restore the data and do your checks. Compare the result with the downtime you said you could live with in step 2. If the real time is longer, you have found the most useful fact in the whole exercise, while there is still time to fix it.
6. Test the copy an attacker cannot reach
Ransomware groups look for backups and try to delete them before they encrypt anything. At least one copy should live away from your production servers, behind separate logins with multi-factor sign in, and ideally in a form that cannot be changed or deleted for a set period. Restore from that copy at least some of the time. A test that only ever uses the convenient local copy tells you nothing about the copy you would rely on after a serious attack.
7. Write it down
A restore test you cannot show someone did not happen as far as an auditor, insurer or new staff member is concerned. Keep a short dated record for every test: which system, which backup copy and its date, where you restored it, who ran it, what you checked, how long it took, what went wrong and what you changed afterwards. Store the records somewhere that survives the loss of the servers they describe. Over a year, that file becomes the evidence that your backups work and the history of every gap you closed.
8. Make it a schedule, not an event
Systems change every month: new software versions, new folders, new staff, new passwords. A restore that worked in spring can fail in autumn. A practical rhythm for a small business is a small restore test every month that rotates through your critical systems, plus a fuller drill once a quarter where you rebuild one important system end to end. Put the dates in a calendar with a named owner, and treat a missed test like any other missed control.
Restore test checklist
- Critical systems listed, each with how much data and how much downtime the business can accept.
- Restore done into a separate, isolated machine or network, never over a live server.
- Encryption keys and passwords needed for the restore were found by the people who would run a real one.
- Files opened, database started and the application used, with the freshness of the newest data noted.
- Total time measured from finding the backup to a usable system, and compared with the target.
- At least one test this quarter used the off-site copy behind separate logins.
- A dated record saved, including problems found and the fixes made.
- Next test dated in the calendar with a named owner.
Next steps
GlassHosting is a managed service and infrastructure provider: we run your servers on our own network, so you can run your business. If you want to know where you stand today, take the free Recovery Proof Score: ten questions and a gap list based on your answers. If you would rather have the tests run for you every month with a dated report, see Provable Recovery. Related reading: backups and snapshots on a Linux VPS and WordPress and WooCommerce backups and uptime.
Host with GlassHosting
Order a plan or browse the knowledgebase for deeper how-tos.