Saturday 17 Jan 2026


How I Archive and Back Up My Photos and Videos Each Year

Every year, I generate a large number of photos and videos that I want to keep safe long term. In 2025 alone, my Pictures and Videos directory grew to 62 GB, consisting mostly of JPEG photos and MP4 videos. Because this data is irreplaceable, I take archiving and backup seriously.

To keep things manageable, I separate my media by year. This makes it easier to reason about what is “active” data versus what is safely archived, and it simplifies both backup and restore processes.

This post outlines how I back up files during the year and how I archive them once the year is complete.

Tools I use:

Backups During the Year

Throughout the year, I treat my photo library as a live dataset.

  • I sync files individually to Backblaze B2, ensuring that new photos and videos are protected soon after they are created.
  • This provides continuous off-site backup and peace of mind in case of disk failure or accidental deletion.

Once the year ends and the photos are fully archived (described below), I delete the Backblaze bucket for that year and start fresh for the next one (for example, 2026). This keeps my backup setup clean and avoids carrying unnecessary historical data in my active sync.

End-of-Year Archiving Process

At the end of the year, I switch from “backup” mode to “archive” mode. This is where long-term durability and integrity really matter.

1. Checksumming Everything

First, I run my own tool, defiant-fg, across the entire directory:

  • Every photo and video is checksummed.
  • The results are stored in a database file.

This gives me a definitive record of what each file should look like and allows me to detect corruption later.

2. Creating Archives with Parity

Next, I use another tool I’ve written, arcK, to package the entire directory:

  • The directory is combined and split into archive parts.
  • Parity data is included so that minor corruption can be repaired.
  • This approach also reduces the number of parts that need to be downloaded during a restore, saving Glacier retrieval costs and bandwidth.

The goal here is not just compression, but resilience.

3. Geographic Redundancy with S3

Once the archives are created, I upload them to two separate S3 buckets:

  • One bucket in the UK (e.g. karl-pictures-uk-part1)
  • One bucket in the USA (e.g. karl-pictures-usa-part2)

This provides regional diversification and protects against large-scale outages or regional failures.

//Example of tree within the S3 bucket. This bucket contains parts numbered 5-10://

├── 2025.10
├── 2025.11
├── 2025.12
├── 2025.5
├── 2025.6
├── 2025.7
├── 2025.8
├── 2025.9
├── 2025.m
└── dfg
    ├── 2025.dfg.db.7z

//And the other bucket contains overlapping parts 0-8://

├── 2025.0
├── 2025.1
├── 2025.2
├── 2025.3
├── 2025.4
├── 2025.5
├── 2025.6
├── 2025.7
├── 2025.8
├── 2025.m
└── dfg
    ├── 2025.dfg.db.7z

Nb: the 2025.m file contains metadata to allow parts to be recombtined. I have copued this twice into each region for backup purposes.

4. Verifying the Archive

Finally, I verify that the archive process itself hasn’t introduced any errors:

  • I extract the 2025 archive using arcK. (I did not download from S3, instead using local copy I had.)
  • I run defiant-fg again on the extracted files.

The results show no errors, confirming that the files were restored perfectly and that no corruption or loss occurred.

This final verification step is critical as it proves that the entire workflow works end-to-end.

Conclusion

Photos and videos are often some of the most valuable data we own, yet they’re surprisingly easy to lose through hardware failure, human error or silent corruption. By separating backups from long-term archives, using checksums, adding parity, and storing data across multiple regions, I’ve built a process that prioritizes integrity, durability and recoverability.

The key takeaway is that a backup isn’t complete until it’s verified. Testing restores and validating checksums may take extra time, but they turn a hopeful strategy into a proven one. For me, that confidence is well worth the effort.


Backlinks:
index
Journal:Index
Journal:2026:01