Jump to content
  • 0

Advice Wanted: Considering switch from RAID mirror to DrivePool


Ned

Question

Hello all,

I have a Windows 2012R2 server with a 4TB RAID1 mirror (2 x 4TB HDD, using basic Windows software RAID, not storage spaces - can't boot from storage spaces pool) and some other non-mirrored data disks (1x 8TB, 2 x 3TB etc.)

This is configured with the OS and "Important Data" on the 4TB mirror, and "Less Important Data" on the other non-redundant disks.

The 4TB mirror is now nearly full, so I intend to either replace this mirror with a larger mirror, or replace this mirror with two larger non-mirrored drives, and use DrivePool to duplicate some data across them.

I'm evaluating whether the greater flexibility of using DrivePool is worth moving to that, instead of using a basic mirror as I am currently.

Current config is this:
C: - redundant bootable partition mirrored on Disk 1 and Disk 2
D: - redundant "Important Data" partition mirrored on Disk 1 and Disk 2, and backed up by CrashPlan
E:, F:, etc. - non-redundant data partitions for "Less Important Data" on Disks 3, 4, etc.

If I moved to DrivePool I guess the configuration would be this:

C: - non-redundant bootable partition on Disk1
D: - drivepool "Important Data" pool duplicated on Disk1 and Disk2, and backed up by CrashPlan
E: - drivepool "Less Important Data" pool spread across Disks 3, 4 etc.

- or -

C: - non-redundant bootable partition on Disk1
D: - drivepool pool spread across all disks, with "Important Data" folders set to only be stored and also duplicated on Disk1 and Disk2, and backed up by CrashPlan (or something similar using hierarchical pools)

I have a few questions about this:

  1. Does it make sense to use DrivePool in this scenario, or should I just stick to using a normal RAID mirror?
  2. Will DrivePool handle real-time duplication of large in-use files such as 127GB VM VHD's, and if so, is there a write-performance decrease, compared to a software mirror?
  3. All volumes use Windows' data-deduplication. Will this continue to work with DrivePool? I understand the pool drive itself cannot have deduplication enabled, but will the drive/s storing the pool continue to deduplicate the pool data correctly?
    1. Related to this, can DrivePool be configured in such a way that if I downloaded a file, then copied that file to multiple different folders, it would most likely store those all on one physical disk so that it can be deduplicated by the OS?
  4. CrashPlan is used to backup some data. Can this be used to backup the pool drive itself (it needs to receive update notifications from the filesystem)?
    1. I believe it also uses VSS to backup in-use files, but I as this is mostly static data storage I think files should not be in use so I may be able to live without that.
    2. Alternatively, I could backup the data on the underlying disks themselves?
  5. Are there any issues or caveats I haven't thought of here?
    1. How does DrivePool handle long-path issues? Server 2012R2 doesn't have long-path support and I do occasionally run into path-length issues. I can only assume this is even worse when pooled data is stored in a folder, effectively increasing the path-length of every file?

       

Thanks!

Link to comment
Share on other sites

3 answers to this question

Recommended Posts

  • 0

1. I think in this case it comes down to your own personal preferences.

2. Personally I wouldn't (the VMs I have in my pool are kept in a folder set to zero duplication), but if I did I'd definitely test it thoroughly before committing to it.

3. The current version of DP should detect dedupe automatically, but to make sure check that "Bypass file system filters" is absent or at least unticked as a per-pool option under "Manage Pool -> Performance". As I understand it however, dedupe will still be per physical volume so you might not see much benefit?
- 3.1 ... maybe? You'll need to investigate whether a custom File Placement rule could provide what you're wanting.

4. Can't answer this one, depends how well CrashPlan plays with virtual drives?
- 4.1 Yeah, I don't think VSS would work.
- 4.2 Unless you did that, of course.

5. Given your setup, you might consider using mirrors for duplication and drivepool for pooling your mirrors?
- 5.1 as I understand it, DP has its own internal support for long path, so - at least as far as the physical *:\PoolPart.* folders storing the pooled data are concerned - you should only have to worry if you have programs/processes that are bypassing DrivePool to talk to the physical drive(s) directly. Windows may of course still run into problems if P:\* itself (where P is your virtual pool drive letter) is longer than 256 characters.

Link to comment
Share on other sites

  • 0

Thanks.

One of the main reasons I was considering switching to DrivePool was for the greater flexibility - currently as it's quite a small system, I have multiple VM's and the OS all on a single mirror - this is obviously bad for performance. I figured with DrivePool I may be able to put the busier VM's on a different disk to each other and to the OS, which should improve performance.

However I've just found out that you can't use dynamic disks in DrivePool (why? their explanation of the disks being more complex makes no sense - surely drivepool operates at the volume level, so who cares about what's below that?) so that sort of shoots my plans in the foot... my mirrors are dynamic disks (it's the only way to do so in Windows software RAID - Storage Spaces doesn't support booting), which means they won't work in drivepool apparently, and also makes it nearly impossible to easily convert from a dynamic disk based system to a drivepool based system without a huge amount of work and copying and temporary spare storage space.

Is there any way to get DrivePool to work on volumes on dynamic disks?

Link to comment
Share on other sites

  • 0

Sorry, I'd forgotten that DP doesn't allow dynamic disks (it's been years since I've had to deal with them). I did find answers here and here as to why in another thread; the TLDR is apparently dynamic disks aren't supported because they (the Windows API) make it difficult and thus slow for DrivePool to ensure any duplicate files are actually on separate physical disks, and customers don't like slow: to quote, "if your product is dirt slow, people will move on. ESPECIALLY when it comes to storage."

Re converting dynamic disks to basic disks there are some third-party partition managers that can do it without having to erase the content, so you might find a trial version to do the job as a one-off? Of course I'd make sure to have current backups just in case.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Answer this question...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...