FM Meets - Hours to Seconds: Rebuilding Our FMS Offsite Backup Pipeline

For years, our offsite FileMaker Server backups relied on rsync — simple to set up, but painfully slow on servers with large numbers of external container files. The real bottleneck wasn’t storage size; most of our servers run 5–50GB of container data, with our worst case around 100GB. It was file count: rsync has to walk and compare tens of thousands of individual files to figure out what changed, and a server with less data but more files can take longer to sync than one with far more data in a handful of large containers. This session opens with a quick audience check on how everyone else is handling offsite backups, before digging into rsync’s limitations: no atomicity, no deduplication, and no way to catch silent data corruption. The fix came from rethinking the problem at the filesystem level: instead of walking and comparing files one by one, advanced filesystems like ZFS take atomic, block-level snapshots of the entire dataset and replicate only the changed blocks — regardless of how many individual files changed. We’ll walk through what that setup looked like in practice, the real-world speed gains, and the pitfalls we hit along the way. Expect concrete before-and-after numbers, a live demo (with a recorded backup demo in case things go sideways), and practical takeaways any FileMaker admin can apply to their own backup strategy.