Lessons from Migrating 9TB of File Shares to FSx

Migrating a Windows file server sounds straightforward until you're staring at 9TB of data across 14 shares and trying to work out what's actually worth moving. This is what I learned doing exactly that — moving a legacy EC2-hosted Windows file server to FSx for Windows File Server, with a detour through S3 Glacier for the data nobody was using. Start with discovery, not migration The temptation is to spin up FSx, robocopy everything across, and call it done. Resist that. You'll end up paying FSx prices for terabytes of data that hasn't been touched in years. I wrote a PowerShell script to scan every share and classify files by age. This immediately surfaced that a significant portion of the data was cold — files that hadn't been written to in over two years. The LastAccessTime trap Here's the gotcha that cost me a day: the server had DisableLastAccess set to 1. This is a common Windows performance optimisation, but it means LastAccessTime is unreliable — it wasn't being updated when f