This add-in will let you add multiple hard drives into one combined storage pool on the Windows Home Server. You can create folder shares on this storage pool, and each folder on the pool can either be duplicated or not duplicated. Files placed in duplicated folders will be protected against single drive failure by being placed onto two separate hard drives.
- Combine your hard drives into one big storage pool.
- Create regular or duplicated folders on the pool. Duplicated folders protect your files against single drive failure.
- Shows up as a standard drive letter, so seamlessly works with client computer backups, media streaming and HomeGroups.
- Add and remove any drive at any time, even with existing non-pooled data. Existing data is not modified.
- All your data is stored in standard NTFS files. You can always get at your files even if you don’t have access to the DrivePool software.
Windows Home Server Integration
- Deeply integrated with the Windows Home Server.
- DrivePool folders are managed from the Windows Home Server Dashboard.
- Support for remote web access and media streaming.
- Support for putting the special Client Computer Backups folder on the pool, giving you unlimited storage space to backup all your computers.
- Windows Home Server folder permissions using the standard folder properties dialog.
- Easily move your existing Windows Home Server folders onto the pool using the standard move folder wizard.
- Support for the Windows Home Server installer means that you can install the add-in from any connected computer by simply running it.
- The pool is self-contained.
- If your computer crashes, connect your pooled drives to another computer and access your pool instantly.
- Instantly access your pool after an operating system re-install.
- Duplicated files remain readable, even in the face of bad sectors or a broken hard drive.
- DrivePool communicates with the NTFS file system driver in the kernel directly, for the fastest speeds possible.
- When a duplicated file is being written to two separate disks, every write happens in parallel in order to minimize the impact on performance.
- Directories are listed by directly querying all the disks in parallel.
- Fast algorithm capable of handling many tens of thousands of files in a single directory without slowing down.
- Results are buffered into a master list, and as the master list is being built the results are returned to the application.
I’ve been using this add-in for quite some time now, and it’s been performing somewhat well. There has been having some hickups forcing me to reboot the server periodically to ensure that file access could be made at any time, but other than that, performance has been noticably better than I would have expected.
Now, Milestone 4 has been released. This changed the initial view that I saw at first, and it got me very happy. I think this is a good update and I can initially notice some very nice performance increases – changes.txt after this screenshot.
In short, if you’re using Windows Home Server 2011 and want the same storage flexibility as you did with Windows Home Server v1 where you easily could add a USB drive as a temporary fix for short storage while that 10kRPM disk is being delivered – you can and should get this add-in. Currently in beta, but stable enough to trust from where I’m seeing this.
* Fix for keeping integrated mode enabled after updating.
* [D] Fixed rename and overwrite issue. Was not allowing directory merging among other things.
* [D] Fixed compatibility issue with volume shadow copy service (for non-pooled drives).
It is a good idea to reboot BEFORE upgrading to this version, just in case VSS is locked up.
If VSS is locked up while upgrading, the upgrade will lock up:
In that case, Reboot -> Dashboard -> Remove DrivePool -> Reboot -> Install latest build.
* Fixed M4 migration issue causing "Access denied" error.
* [D] Workaround for issue in rdbss (bug?) that denies a flush on a read-only handle.
This incorrectly returns access denied errors for non-write operations.
This only affects files accessed using a UNC share path, locally on the server.
Accessing shares from remote or from the local drive letter does not trigger this problem.
* [D] When cloning directories, clone the owner properly.
* [D] Improved performance / accuracy of file size tracking.
* [D] Better support for empty pool. Treat as a disk that has 0 bytes free.
* Fixed Dashboard crash on missing disk.
* Added BitLocker support for pool parts (disabled by default - see .config).
* Fixed null reference when removing drives from an unmounted pool.
* Fixed pool condition UI bar getting stuck at 0.
* Fixed deadlock when 2 background services were attempting to start at the exact same time. Did not stall I/O.
* When removing a drive, show it as such, instead of missing disk.
* When removing a drive, report duplication warnings.
* Do not automatically duplicate when encountering an incomplete file if there are duplication warnings that need resolutions.
Give the user a chance to view the warnings and resolve the issue.
This does not affect nightly duplication runs (when there are unduplicated files present).
This does not affect instant duplication runs when moving a file from a duplicated folder to a non-duplicated folder and vice versa.
This does not affect duplication runs when adding a non-empty pooled disk, et al.
* Do not list system reserved volume in Pool tab.
* Volume name is now respected by DrivePool tabs.
* [D] Pool drive can now have a volume name.
* [D] Changed driver load order to address disappearing shares on reboot.
* [D] Fixed timestamp issue in disk driver. Was preventing installation from going through on some systems.
* Added new Drives sub-tab under the DrivePool tab. Shows disk space usage on all pooled drives.
* Tweaked folder details presenter.
* [D] Changed pool part initialization / refresh to be synchronous.
* [D] Added support for undocumented directory listing file system behavior having to do with indexed requests.
Managed to keep the same parallelized performance aspect of the algorithm, while adding deterministic ordering of results.
Indexed queries are not as fast as non-indexed queries, but they are rarer (E.g. Triggered by: XP, OS X, boxee, MS-DOS applications, ...),
and are only issued for large directory listings.
* [D] Tightened up the directory listing algorithm code.
* Fixed installation issue, where if the previous M4 install failed in the middle, future installs would not work.
* [D] Fixed directory enumeration issue that was causing the WHS client backup engine to report damage in the backup database.
Note that, when reproducing this issue, the backup database files were not actually damaged.
* [D] Fixed simple rename crash (E.g. when renaming from OS X).
* [D] Disk buffer fixes.
* Attempted to fix folder re-share on migration (unable to reproduce).
* [D] Fixed directory listing. Sometimes it would incorrectly tell the caller that there are no more files.
* [D] Changed disk buffer access one more time to use exception handling instead of relying on WDF.
* [D] Cleanup success code is now consistent with IRP.
* [D] Fixed crash in handling lock control.
* [D] Fixed virtual physical disk buffer access (was causing a number of crashes).
This buffer is really meaningless, as we don't do much with it.
* [D] Fixed rename + replace issue. If an application requested to rename / move a file and replace the destination in one
operation, then there was a chance that the destination file would not be replaced properly. The duplication engine
would detect this later as a file conflict and offer to fix the situation be deleting the older file part.
* [D] Fixed crash when reading data.
* [D] Fixed crash when FAT32 (or other file systems) volumes are present or plugged in (such as USB thumb drives).
There was a bug in the file system identification code.
* [D] Brand new file system driver, CoveFS.
* Not based on old user mode / kernel mode hybrid FS approach.
* 100% native solution, on par with the built-in file systems.
* Built from scratch specifically for pooling.
* Does not rely on any user mode service / program to function.
* Mimics NTFS for many operations.
* Direct file I/O pathways for ALL operations with Fast I/O support.
* Fast directory listing using parallelized algorithm.
* Pool loads at boot time, like regular file system over a virtual disk.
* [D] Brand new virtual disk drive.
* Pool volume sits on this virtual drive.
* You can mount drive letters to it, etc... Just like any other disk.
* Shows up in computer / server management.
* Full plug and play / power support.
* Pool is now mounted using a drive letter. You can work with it, like any other drive.
* Integrated UI.
* Integrates elements of DrivePool into the default "Server Folders and Hard Drives" tab.
* Allows you to fully manage your pool from the standard tabs.
* Enhances the detail panes with information from the Windows Indexing service and DrivePool.
* This UI mode is optional and can be turned off, which takes you to a M3-like UI.
* DrivePool overview tab.
* Displays all the statistics for your pool in one place.
* Links to various other related areas of the Dashboard.
* Can be completely hidden in integrated mode for a native look and feel.
* DrivePool settings.
* Integrated into standard Dashboard settings UI.
* Standard built-in wizards for add / move / delete folder.
* Best compatability with the existing Microsoft code.
* Full compatibility with media streaming / HomeGroups out of the box.
* Client computer backups folder can be moved to the pool using the standard wizard.
* Background duplication.
* Changes to file duplication happen in the background.
* User is not forced to wait when setting the duplication status for a folder.
* Files are duplicated using Background I/O (does not interfere with other disk access).
* Pool condition percent replaces balance ratio.
* Pool condition indicates balance ratio and duplication status in one place.
* Allows you to manually initiate balancing / duplication actions.
* Pool can function in 2 different duplication modes.
* Same idea as M3.
* Files are duplicated as they are written.
* Writes happen to 2 separate hard drives, in parallel.
* Duplicate later
* Files are always written to one hard drive.
* Files that need to be duplicated are copied to the second drive during the nightly maintenance cycle.
* This carries some restrictions. Naturally, files that are in use during the maintenance cycle cannot be duplicated.
* The duplication maintenance cycle happens only once a day, at night, and only if this mode is enabled.
* Updated icon :)
* All the other goodness from M3 is mostly still here.
* Real-time folder size tracking.
* Real-time pool balance ratio to let the algorithm know if balancing is "necessary".
* Real-time pool usage pie chart.
* Add / remove disks to the pool with existing data on them without altering the existing data.
* Pooled files are stored as standard NTFS files on each pooled disk.
^^^ 1.0 BETA M4 ^^^
* [D] Fixed crash on enumerating disks.