Btrfs was merged into the Linux kernel in 2009, arriving with bold promises—and, let's be honest, a reputation for instability. I first tried it on my laptop in 2011. It wiped my data. Twice. On the bright side, it taught me the value of backups.

Fast forward to 2025: btrfs is no longer the experimental filesystem of the past. It's stable, mature, feature-rich, and fully part of the Linux kernel. But old reputations die hard. Even today, Google Cloud Platform doesn’t officially support it—not because of technical shortcomings, but because customer demand hasn’t pushed the issue.

At Chronosphere, we decided to take a fresh look. After months of evaluation and testing, we migrated petabytes of customer data across thousands of disks to btrfs. This talk is our story: why we made the leap, what we learned along the way, and how we’re helping bring btrfs into wider enterprise adoption—including working with Google to support it natively.

I'll share the decision-making process, key performance and reliability insights, and the quirks you only discover when running btrfs at scale. Whether you're btrfs-curious or just love a good ops tale, you'll walk away with real-world takeaways—and maybe a newfound respect for this once-maligned filesystem.

If you know what NTFS, ext4, or ZFS are, you’re ready for this journey.