Why it never gets done
The job stalls when someone asks what is on it and nobody knows. So it stays on.
Keep all of it. Deciding what matters costs more than storing it.
The order to work in
- Measure it. Total size, file count, and the date of the newest file in each top-level folder. Most of the tree stopped changing years ago.
- Find the live corners. Anything changed in the last year belongs in your working system, not in an archive. It is usually a small fraction.
- Copy everything else off, checked. Every file compared with the original by checksum, with a report you keep.
- Make it searchable, by name and by the words inside documents.
- Leave a sign. A note where the old share was, saying where it went and how to get in.
- Wait one cycle. Leave the server on but read-only for a month, over a month-end or quarter-end, so anything seasonal surfaces.
- Then switch it off, keep the disks for a month, and dispose of them properly with a certificate.
Problems to expect
- Long paths and odd characters. Decades of Mac and Windows users leave filenames that break naive copies. The copy must report what it could not take.
- Permissions nobody understands. Do not recreate twelve years of access rules. Decide who should see it now.
- Shortcuts and aliases pointing at drive letters that will not exist. Knowing which ones break is better than discovering later.
- The one database somebody is still using, in a folder called Old. Changed-date reporting finds it before you pull the plug.
- Backups of the server that you keep paying for afterwards. Cancel them deliberately, once the archive is replicated.
What proof should look like
The proof is a file you can show an auditor or an insurer: every file that was on the server, its size, its checksum, and confirmation that the copy matches.
How Shed does it
The contents come across whole, folder structure intact, and every file is checked against the original by hash. What lands is searchable by name and inside documents, previewable in a browser, and organised the way your team remembers. It is replicated across our nodes, and the report stays with the data.
Tell us roughly how big it is and we will size it and give you a date.