Satoru Takeuchi of Fujitsu presented at last month's LinuxCon Europe event about Btrfs in its current state as well as future prospects. In terms of Fujitsu's stake in Btrfs, they have been working on Btrfs for mission critical systems since 2010. Fujitsu's mission critical system requirements include a high level of robustness with regard to no crashes, error detection, repair and recovery. Fujitsu MC systems also need high availability with extremely limited maintenance windows. Btrfs fits these requirements, at least in theory, and Fujitsu has been investing in improving the file-system upstream.
The information in the presentation isn't entirely new if you're an ardent Phoronix reader with our hundreds of articles about Btrfs and dozens of benchmarks over the years. Promoted was Btrfs support for sub-volumes, snapshots, transparent compression, data/meta-data check-sums, Copy-on-Write updates, etc. Recent kernel releases have worked on improved error handling, inode properties, offline deduplication, and improving performance. Fujitsu contributes new features not just to the Btrfs kernel module but also to the user-space Btrfs tools too.
Fujitsu's Takeuchi declares Btrfs ready to use as long as you're not using it in a RAID 5/6 array, the performance and stability should be good, and the features are good. Running Btrfs in RAID 1 and RAID 10 are deemed the best choices right now. If you missed it, in the past few days I posted a number of Linux RAID 0/1/5/6/10 benchmarks using a variety of file-systems on Ubuntu.
Future changes for Btrfs include scrub/replace sub-commands for RAID 5/6, better robustness and performance, improved documentation, etc.
For those that weren't at LinuxCon 2014 in Germany, you can find the slides in PDF form.