1

Assuming a simple, fully-provisioned two-way mirror Storage Spaces setup on Windows 10/11, how does it behave when one of the drives gets faulty:

  • One of the drives cannot read some sectors, and SMART's uncorrectable error count goes up. How does SS know it should instead return data from the other copy on another drive?
  • SMART status stays normal but it returns erroneous values (due to bit rot for example)?

In general, how does a mirrored SS know which drive (copy) to "trust"?

Especially considering it may be doing performance optimizations when reading and read from both drives in parallel (albeit different chunks of data)?

In a 3+-way mirror it would be even more interesting to see, since in this case a some kind of consensus algorithm might be used (i.e. majority rules, so it would use the version of the data which has the most matches). However this goes head-to-head with the notion of performance when reading. If it were doing such a consensus validation, then the read speed would be the same (if not slower) than not having any kind of redundancy at all.

Improve this question

1 Answer 1

Reset to default
1

I'm not going to pretend that I'm an expert to this, but I'm interested in this topic and I made some reserach. I will share some link here.

According to the documentation of ReFS file system: https://learn.microsoft.com/en-us/windows-server/storage/refs/refs-overview

Resiliency features are related to the file system itself. My understanding is that if you use NTFS on top of Storage Spaces, you don't have that kind of support.

Especially interesting are ReFS integrity streams. ReFS uses metadata, including checksums, to verify if the data is corrupted or not. When detects corruption, it will try to recover the data from the seond copy. As far as I know, Integrity stream are not enabled by default, but linked documentation shows how to enable it.

ReFS look quite nice, unfortunately, since some time, it is available only on Win11 Pro for workstations: https://www.microsoft.com/en-us/windows/business/compare-windows-11

On the other hand, storage spaces itself provides a Repair-Volume command: https://learn.microsoft.com/en-us/powershell/module/storage/repair-volume?view=windowsserver2019-ps

However, if you look at documentation, it is not very extensive in term what it does.

Improve this answer
3
  • Thank you for your information. I was more interested in the Storage Space part itself, which is File-System agnostic. In fact I have one such space using VeraCrypt, which until mounted is seen as RAW partition, and yet, when I was doing some tests and removed one of the drives, while having had another one updated, it correctly picked up the changes and synced the drives (repaired volume). So, regardless of File System, it must be having some kind of procedure to determine which copy to use when. I was interested in the details of that.
    – Fit Nerd
    Commented Sep 29, 2023 at 15:03
  • Yes, that's the part I'm missing to. Advertising of ReFS makes impression that you need that filesystem to get resiliency. I need to dive more into documentation of Storage Spaces itself. Also, because I'm considering investing in ReFS.
    – b00rt00s
    Commented Sep 29, 2023 at 15:07
  • I understand. In case you find anything related to this, please share your info!
    – Fit Nerd
    Commented Sep 29, 2023 at 15:28

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .