I'm curious if anyone has been using the new ReFS deduplication feature and what your experiences have been like. I came across some information about it, and I'm particularly interested in how it impacts storage pools where ReFS is used to store VHDX files. I've been using ReFS for everything except the hypervisor's boot volume and it's been pretty stable overall, with some nice surprises. Notably, I've been using it for VM's NTFS boot VHDXs and I love the instant functionality, especially with Server 2025's native block cloning feature. However, I've started to think that deduplication might just be addressing symptoms of poor data management practices instead of fixing underlying issues. It seems suited for specific use cases, and if you're in that niche, you'll know it.
2 Answers
I've faced several problems with ReFS deduplication. The performance issues are severe—it's painfully slow, almost unusable at times. I've had complete ReFS volumes go offline or show up in RAW format without any way to recover them. Now, I only trust ReFS for certain uses—like CSV for Hyper-V disk images. Other than that, I'm all about NTFS and keeping good backups. Deduplication might work fine, but the performance hit just isn’t worth it. If you do go for it, back everything up and make sure those backups are solid!
Honestly, I've had quite a few issues with ReFS. It wouldn't release space after deleting files, and there were instances of file corruption. For now, I'm sticking with NTFS.
Yeah, it seems like OP missed the rough patches from back in 2016 and 2019 when ReFS just failed at times. Microsoft really hasn't managed to create a reliable file system since NTFS; remember WinFS? They scrapped that before it even launched, it was that bad.