Your Backup Has Never Been Tested: Run the 90-Minute Restore Drill That Turns Hope Into Proof

Your backup is not a backup yet. It is a hypothesis. You believe the drive works. You believe the cloud folder is synced. You believe the files will open on a machine that is not the one that created them. But belief is not a restore. And most people have never actually tested the thing they are counting on.

That is the uncomfortable truth behind a lot of video archives: the backup has passed a vibe check, not a fire drill. The drive lights up. The folder opens. The thumbnail loads. Good enough, right? No. A backup that has never been restored is a hope with a USB cable.

Here is a 90-minute drill you can run this weekend. It will not test every file. It will test the system, which is the part that fails.

What Counts as a Real Restore Test

A real test means copying a file from each backup location onto a clean machine, opening it, and confirming that it is the file you think it is. Not browsing. Not previewing from the cloud. Not clicking play in the same app that manages the backup.

You need a machine that is not your daily driver if possible: a spare laptop, a bootable USB environment, or a clean virtual machine. You need a media player like VLC, a checksum tool, and a plain spreadsheet. You also need a timer, because restore time matters. A backup that takes three days to retrieve is not the same as a backup that takes three hours.

Before the Clock Starts

Gather the credentials for every copy: the local drive, the second drive, the offsite drive, the cloud account. If you need a password, a recovery code, or a hardware key, have it ready. A restore test that stalls because you cannot log in is already telling you something important.

Set up the clean machine. Install VLC. Install a checksum utility or use the command line. Open a spreadsheet with columns for file name, source copy, expected size, hash, playback result, captions, metadata, and notes. Print it if you like. The point is to record failures without relying on memory.

Minutes 0-10: Pick a Random Sample

Do not pick your favorite videos. Pick random ones. Use a random number generator against your catalog, or scroll with your eyes closed and jab. Aim for 10 files across different years, formats, sizes, and sources. Include at least one long video, one phone clip, one file with subtitles, and one file from each backup location.

Write down what you expect before you restore anything. File size. Duration. Resolution. Whether it should have captions. This is the baseline. Without it, you will talk yourself into thinking a broken file is fine.

Minutes 10-30: Restore From Copy A

Start with your primary local copy. Copy the sample files to a fresh folder on the clean machine. Do not open them from the original drive. The whole point is to simulate a restore, not a preview.

Verify the checksum if you have one. If you do not, compare file size and duration, then play the first 30 seconds, the middle, and the last 30 seconds. Check audio sync. Open the captions. Note anything that stutters, freezes, or looks wrong. Write it down even if it seems minor.

Minutes 30-50: Restore From Copy B

Move to the second copy. This might be the offsite drive or the second local drive. Same process. Copy, verify, play. Pay attention to how long it takes to access the copy. If the drive is in a safe deposit box or a relative's closet, you are really testing your retrieval plan, not just the file.

If the second copy is a different format or a different folder structure, note that too. A backup that requires a decoder ring is a backup that will fail when someone else needs to use it.

Minutes 50-70: Restore From the Cloud

Download the sample files from the cloud account. Watch the transfer speed. If the provider throttles you or the files are in an archive tier that needs a restore request, mark the time. That delay is part of your recovery time, whether you like it or not.

Check versioning and deleted-file recovery while you are there. A cloud copy is not much use if a sync error overwrote the good version with a zero-byte file. Confirm that the file you downloaded is the file you expected, not just the file that happens to have the right name.

Minutes 70-85: Check Playback, Subtitles, and Metadata

This is where a lot of archives quietly fail. The video plays, but the audio is out of sync. The captions are missing because the sidecar .srt file was never copied. The creation date is gone. The title and description live in a database that did not make the trip.

Open each file in MediaInfo or ExifTool. Check codec, resolution, duration, and creation date. Open the subtitles separately if they are sidecar files. Confirm that your catalog still points to the right file. If you rely on a folder structure to carry meaning, make sure the structure survived the restore.

Minutes 85-90: Write the Failure List

Do not end with a feeling. End with a list. What broke? What was missing? What took too long? Which credential did you have to hunt down? Which file played on the old machine but not the clean one?

Assign a fix and a date to each item. Replace the failing drive. Add the missing subtitle file. Update the restore notes. Store the credentials in the password manager. Then put the next drill on the calendar. Quarterly is reasonable for a personal archive. Once a year is the bare minimum.

The Questions the Drill Answers

Can you actually restore a file? How long does it take? Do you have the keys? Are captions included? Is the cloud copy synced or just present? Does the drive mount on a machine that is not yours? Can someone else follow your notes if you are unavailable?

Those are not abstract questions. They are the difference between an archive and a collection of storage devices. At SaveVid, we think preservation is a practice, not a purchase. The drill is how you find out whether the practice works.

Run it once. Fix what breaks. Run it again. A backup you have restored from is a backup. Everything else is a hope with a progress bar.