rt/snapraid
1
0
Fork 0
mirror of https://github.com/amadvance/snapraid.git synced 2026-10-26 19:01:57 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Auto update

Browse source
This commit is contained in:
Andrea Mazzoleni 2025-10-14 09:39:51 +02:00
parent e69776f3be
commit 970147fefa
1 changed files with 44 additions and 84 deletions
Download patch file Download diff file Expand all files Collapse all files

128
compare.html
View file

@ -54,16 +54,16 @@
<p>
<span class="bold">SnapRAID</span> is just one of many <span class="bold">RAID-like</span> solutions available for disk arrays.
<span class="bold">SnapRAID</span> is one of many <span class="bold">RAID-like</span> solutions available for disk arrays.
</p>
<p>
Some of the best-known alternatives include:
Some well-known alternatives include:
</p>
<ul>
<li><a href="http://www.lime-technology.com/">unRAID</a>
- A commercial and open-source (GPL2) solution. It uses a modified Linux ReiserFS filesystem with real-time redundancy. Integrity checksums are not supported.
- A commercial and open-source (GPL2) solution. It uses a modified Linux XFS filesystem with real-time redundancy. Integrity checksums are not natively supported.
</li>
<li><s>FlexRAID</s>
- <b>DISCONTINUED</b>. A commercial, proprietary C++/Java application for Windows, with limited Linux support. It offered both snapshot and real-time redundancy, including integrity checksums.
@ -73,46 +73,36 @@ Some of the best-known alternatives include:
</li>
<li><a href="http://en.wikipedia.org/wiki/btrfs">Btrfs</a>
- An open-source (GPL2) Linux filesystem with real-time redundancy. Since Linux 3.9, it supports <a href="http://lwn.net/Articles/536038/">RAID5/6 redundancy</a> and integrity checksums.
Recent issues have been discovered, and the official <a href="https://btrfs.wiki.kernel.org/index.php/RAID56">wiki</a> now warns:
<i>"The parity RAID code has multiple serious data-loss bugs in it. It should not be used for anything other than testing purposes."</i>
Recent issues have been identified, and the official <a href="https://btrfs.wiki.kernel.org/index.php/RAID56">wiki</a> warns:
<i>"The parity RAID code has multiple serious data-loss bugs. It should not be used for anything other than testing purposes."</i>
</li>
<li><a href="https://learn.microsoft.com/en-us/windows-server/storage/storage-spaces/overview">Storage Spaces</a>
- Microsoft’s solution, integrated into Windows 8 and later. Proprietary, with real-time redundancy but without checksumming support. It also has several <a href="http://helgeklein.com/blog/2012/03/windows-8-storage-spaces-bugs-and-design-flaws/">limitations</a>.
- Microsoft’s solution, integrated into Windows 8 and later. Proprietary, with real-time redundancy but no checksum support. It also has several <a href="http://helgeklein.com/blog/2012/03/windows-8-storage-spaces-bugs-and-design-flaws/">limitations</a>.
</li>
</ul>
<p>
The main factor to categorize them is when the redundancy information is updated.
In <span class="bold">real-time</span> solutions the parity data is updated in real-time,
without an explicit action from the user. Like in standard RAID.
In <span class="bold">snapshot</span> solutions the parity data is updated only upon
an explicit request from the user. Like in backups.
The primary factor for categorizing these solutions is when redundancy information is updated.
In <span class="bold">real-time</span> solutions, parity data is updated in real-time
without explicit user action, similar to standard RAID.
In <span class="bold">snapshot</span> solutions, parity data is updated only upon explicit user request, similar to backups.
</p>
<p>
Other important factors are the support of an <span class="bold">integrity</span> checksum
to identify <a href="http://www.zdnet.com/blog/storage/data-corruption-is-worse-than-you-know/191">silent data corruption</a>
and the possibility to <span class="bold">fix it</span>,
if you can <span class="bold">use already filled disk</span>,
if you can <span class="bold">recover your data from not broken disks</span> when you get too many failures to have a full recover,
and if <span class="bold">all your disks are spinning when reading one file</span>.
Other key factors include support for <span class="bold">integrity</span> checksums to detect <a href="http://www.zdnet.com/blog/storage/data-corruption-is-worse-than-you-know/191">silent data corruption</a> and the ability to <span class="bold">fix it</span>, whether you can <span class="bold">use already filled disks</span>, whether you can <span class="bold">recover data from non-failed disks</span> when there are too many failures for full recovery, and whether <span class="bold">all disks spin when reading one file</span>.
</p>
<table class="compare">
<tr><th width="30%"></th><th><span class="bold">SnapRAID</span></th><th>unRAID</th><th>ZFS</th><th>Btrfs</th><th>Storage Spaces</th></tr>
<tr><td><span class="bold">Redundancy model</span><br>Real-time computes parity at real-time like RAID,
snapshot at request time, like Backup.</td>
<tr><td><span class="bold">Redundancy model</span><br>Real-time computes parity in real-time like RAID; snapshot computes parity on request, like backups.</td>
<td>Snapshot</td><td>Real-time</td><td>Real-time<br>Snapshot</td><td>Real-time<br>Snapshot</td><td>Real-time</td></tr>
<tr><td><span class="bold">Integrity</span><br>
If data is validated with some kind of checksum,
and which one is the default.</td>
<td>Yes<br>SpookyHash<br>128 bit</td><td>No <a href="#note8">[8]</a></td><td>Yes<br>fletcher4<br>256 bit</td><td>Yes<br>CRC32C<br>32 bit</td><td>No <a href="#note9">[9]</a></td></tr>
<tr><td><span class="bold">Integrity</span><br>Whether data is validated with a checksum, and the default checksum type.</td>
<td>Yes<br>SpookyHash<br>128-bit</td><td>No <a href="#note8">[8]</a></td><td>Yes<br>Fletcher4<br>256-bit</td><td>Yes<br>CRC32C<br>32-bit</td><td>No <a href="#note9">[9]</a></td></tr>
<tr><td><span class="bold">Fix silent errors</span><br>
If silent errors are detected and fixed before they can propagate into the parity.</td>
<td>Yes</td><td>No <a href="#note1">[1]</a></td></td><td>Yes <a href="#note3">[3]</a></td><td>Yes <a href="#note3">[3]</a></td><td>No</td></tr>
<tr><td><span class="bold">Fix silent errors</span><br>Whether silent errors are detected and fixed before propagating to parity.</td>
<td>Yes</td><td>No <a href="#note1">[1]</a></td><td>Yes <a href="#note3">[3]</a></td><td>Yes <a href="#note3">[3]</a></td><td>No</td></tr>
<!---
Silent error discussion in UnRAID Forum:
@ -136,10 +126,8 @@ http://forum.flexraid.com/index.php/topic,1753.msg12512.html#msg12512
http://www.openegg.org/forums/posts/list/727.page
-->
<tr><td><span class="bold">Number of failures</span><br>
How many disk failures are supported? 1 is for RAID5, 2 for RAID6.
</td>
<td>1 2 3 4 5 6</td><td>1 2</td><td>1 2 3</td><td>1 2</td><td>1 2</td></tr>
<tr><td><span class="bold">Number of failures</span><br>How many disk failures are supported? 1 for RAID5, 2 for RAID6.</td>
<td>1, 2, 3, 4, 5, 6</td><td>1, 2</td><td>1, 2, 3</td><td>1, 2</td><td>1, 2</td></tr>
<!---
unRAID 6.2 adds Dual parityvs ZFS
@ -148,29 +136,28 @@ New Storage Spaces adds Dual Parity if you have at least 7 disks
https://blogs.technet.microsoft.com/tip_of_the_day/2014/07/23/tip-of-the-day-dual-parity-storage-spaces/
-->
<tr><td><span class="bold">More disk failures</span><br>If more disks fail than the supported redundancy model,
is it possible to recover the data in the not failed disks?</td>
<tr><td><span class="bold">More disk failures</span><br>If more disks fail than the supported redundancy model, can data on non-failed disks be recovered?</td>
<td>Yes</td><td>Yes</td><td>No</td><td>No</td><td>No</td></tr>
<tr><td><span class="bold">Power/Noise</span><br>How many disks are spinning when reading a single file?</td>
<tr><td><span class="bold">Power/Noise</span><br>How many disks spin when reading a single file?</td>
<td>One</td><td>One</td><td>All</td><td>All</td><td>All</td></tr>
<tr><td><span class="bold">Filled disks</span><br>Can you start with already filled disks?</td>
<td>Yes</td><td>Partial<a href="#note4">[4]</a></td><td>No</td><td>No</td><td>No</td></tr>
<td>Yes</td><td>Partial <a href="#note4">[4]</a></td><td>No</td><td>No</td><td>No</td></tr>
<tr><td><span class="bold">Add disks</span><br>Can you add disks at later time?</td>
<td>Yes</td><td>Yes</td><td>No<a href="#note5">[5]</a></td><td>Yes</td><td>Yes</td></tr>
<tr><td><span class="bold">Add disks</span><br>Can disks be added later?</td>
<td>Yes</td><td>Yes</td><td>No <a href="#note5">[5]</a></td><td>Yes</td><td>Yes</td></tr>
<!--
ZFS does't allow to add new disks to a raidz pool
http://serverfault.com/questions/190207/how-can-i-add-one-disk-to-an-existing-raidz-zpool
-->
<tr><td><span class="bold">Operating system</span><br>Which OS is supported?</td>
<td>Linux<br>Windows<br>Mac&nbsp;OS&nbsp;X<br>OpenIndiana<br>Solaris<br>BSD</td><td>Linux</td><td>Linux <a href="#note6">[6]</a><br>Mac&nbsp;OS&nbsp;X<br>OpenIndiana<br>Solaris<br>BSD</td><td>Linux</td><td>Windows</td></tr>
<tr><td><span class="bold">Operating system</span><br>Which operating systems are supported?</td>
<td>Linux<br>Windows<br>macOS<br>OpenIndiana<br>Solaris<br>BSD</td><td>Linux</td><td>Linux <a href="#note6">[6]</a><br>macOS<br>OpenIndiana<br>Solaris<br>BSD</td><td>Linux</td><td>Windows</td></tr>
<tr><td><span class="bold">Age</span><br>The year of the first official release supporting at least RAID5 redundancy.</td>
<td>2011</td><td>2005</td><td>2005</td><td>not yet stable <a href="#note11">[11]</a></td><td>2012</td></tr>
<tr><td><span class="bold">Age</span><br>Year of the first official release supporting at least RAID5 redundancy.</td>
<td>2011</td><td>2005</td><td>2005</td><td>Not yet stable <a href="#note11">[11]</a></td><td>2012</td></tr>
<!---
ZFS 2009 is the data of the "Triple-Parity RAID-Z" article by Adam Leventhal
@ -180,89 +167,63 @@ RAID 1987 is the date of the paper:
http://www.eecs.berkeley.edu/Pubs/TechRpts/1987/5853.html
-->
<tr><td><span class="bold">License/Price</span><br>
Software license and price.
</td>
<td>Open Source GPL3<br>Free</td><td>Open Source GPL2<br>59$/129$</td><td>Open Source CDDL<br>Free</td><td>Open Source GPL2<br>Free</td><td>Proprietary<br>Windows 8/10</td></tr>
<tr><td><span class="bold">License/Price</span><br>Software license and cost.</td>
<td>Open Source GPL3<br>Free</td><td>Open Source GPL2<br>$59/$129</td><td>Open Source CDDL<br>Free</td><td>Open Source GPL2<br>Free</td><td>Proprietary<br>Included with Windows 8/10</td></tr>
<tr><td><span class="bold">Interface</span><br>Which interface is provided? GUI or command line?</td>
<td>
command line<br>or <a href="https://github.com/Smurf-IV/Elucidate">Elucidate</a> GUI,
or plugin for for <a href="https://www.openmediavault.org/">OpenMediaVault</a>
</td><td>command line<br>GUI</td><td>command line<br>GUI/Plugins <a href="#note7">[7]</a></td><td>command line<br>GUI<a href="#note10">[10]</a></td><td>GUI</td></tr>
<td>Command line<br><a href="https://github.com/Smurf-IV/Elucidate">Elucidate</a> GUI,<br>or plugin for <a href="https://www.openmediavault.org/">OpenMediaVault</a></td>
<td>Command line<br>GUI</td><td>Command line<br>GUI/Plugins <a href="#note7">[7]</a></td><td>Command line<br>GUI <a href="#note10">[10]</a></td><td>GUI</td></tr>
</table>
<h3>Notes</h3>
<p>
<a id="note1">[1]</a> - unRAID doesn't have any kind of checksum, and it just ignores silent errors.
Even worse, if a parity error is detected as result of a silent error in the data,
the parity is automatically recomputed, making impossible to recover the silent error, even manually.
<a id="note1">[1]</a> - unRAID lacks native checksums and ignores silent errors. If a parity error results from a silent data error, the parity is automatically recomputed, making it impossible to recover the silent error, even manually.
</p>
<p>
<a id="note3">[3]</a> - ZFS and Btrfs provide bit-rot protection comparable to SnapRAID, always verifying data before use.
In this regard, all three solutions represent the state of the art.
<a id="note3">[3]</a> - ZFS and Btrfs provide bit-rot protection comparable to SnapRAID, always verifying data before use. These three solutions represent the state of the art.
<br>
One drawback of ZFS is that the default <a href="http://en.wikipedia.org/wiki/Fletcher%27s_checksum">Fletcher</a> checksum prioritizes speed over reliability.
Even though it is a 256-bit checksum, Fletcher is generally considered weak. Similarly, Btrfs uses CRC32C by default, which is strong, but ony 32 bits.
A better alternative is "edonr" for ZFS and "xxhash" for Btrfs.
In contrast, SnapRAID uses the 128-bit <a href="http://burtleburtle.net/bob/hash/spooky.html">SpookyHash</a>,
which provides state-of-the-art checksum quality without compromising speed.
One drawback of ZFS is that its default <a href="http://en.wikipedia.org/wiki/Fletcher%27s_checksum">Fletcher</a> checksum prioritizes speed over reliability. Although it is a 256-bit checksum, Fletcher is considered weaker than alternatives. Similarly, Btrfs uses CRC32C by default, which is strong but only 32-bit. Better alternatives are "edonr" for ZFS and "xxhash" for Btrfs. In contrast, SnapRAID uses the 128-bit <a href="http://burtleburtle.net/bob/hash/spooky.html">SpookyHash</a>, which offers state-of-the-art checksum quality without sacrificing speed.
<br>
ZFS also employs a sub-optimal RAID-Z3 algorithm that requires roughly twice the computation compared to SnapRAID’s z-parity.
ZFS also employs a suboptimal RAID-Z3 algorithm that requires roughly twice the computation compared to SnapRAID’s z-parity.
</p>
<p>
<a id="note4">[4]</a> - unRAID allows to use filled disks but only if they are already
formatted with the XFS or Btrfs filesystems.
But not ext4 or NTFS, the two most common filesystems used in Linux and Windows.
<a id="note4">[4]</a> - unRAID supports filled disks only if they are formatted with XFS or Btrfs filesystems, but not ext4 or NTFS, the two most common filesystems for Linux and Windows.
</p>
<p>
<a id="note5">[5]</a> - ZFS allows to add a new vdev at the pool, but not a disk to a vdev.
See <a href="http://louwrentius.com/the-hidden-cost-of-using-zfs-for-your-home-nas.html">The 'Hidden' Cost of Using ZFS for Your Home NAS</a>.
<a id="note5">[5]</a> - ZFS allows adding a new vdev to a pool, but not a disk to an existing vdev. See <a href="http://louwrentius.com/the-hidden-cost-of-using-zfs-for-your-home-nas.html">The 'Hidden' Cost of Using ZFS for Your Home NAS</a>.
</p>
<p>
<a id="note6">[6]</a> - ZFS is also available in Linux at kernel level using <a href="http://zfsonlinux.org/">ZFS on Linux</a>.
<a id="note6">[6]</a> - ZFS is available on Linux at the kernel level via <a href="http://zfsonlinux.org/">ZFS on Linux</a>.
</p>
<p>
<a id="note7">[7]</a> - ZFS has external GUIs, like <a href="http://www.napp-it.org/">Napp-it</a>
and plugins for <a href="http://www.freenas.org/">FreeNAS</a> and <a href="http://www.nas4free.org/">NAS4Free</a>.
<a id="note7">[7]</a> - ZFS supports external GUIs, such as <a href="http://www.napp-it.org/">Napp-it</a>, and plugins for <a href="http://www.freenas.org/">FreeNAS</a> and <a href="http://www.nas4free.org/">NAS4Free</a>.
</p>
<p>
<a id="note8">[8]</a> - unRAID can have integrity checksum using a plugin lik
<a href="https://forums.lime-technology.com/topic/43290-dynamix-file-integrity-plugin/">Dynamix File Integrity</a>,
<a href="https://forums.lime-technology.com/topic/41708-checksum-suite/">Checksum Suite</a> or
<a href="https://forums.lime-technology.com/topic/35615-bunker-yet-another-utility-for-file-integrity-checks/">bunker</a>
but they are all independent at the parity processing and not used to help the recovering process.
For example, when recovering a single failed disk with double parity,
the checksum can be used to identify additional silent errors, and still be able to recover.
<a id="note8">[8]</a> - unRAID can use integrity checksums via plugins like <a href="https://forums.lime-technology.com/topic/43290-dynamix-file-integrity-plugin/">Dynamix File Integrity</a>, <a href="https://forums.lime-technology.com/topic/41708-checksum-suite/">Checksum Suite</a>, or <a href="https://forums.lime-technology.com/topic/35615-bunker-yet-another-utility-for-file-integrity-checks/">bunker</a>. However, these are independent of parity processing and not used to aid recovery. For example, when recovering a single failed disk with dual parity, checksums can identify additional silent errors to facilitate recovery.
</p>
<p>
<a id="note9">[9]</a> - Storage Spaces can be used with ReFS that support integrity checksum, but not in the Parity mode:
<a href="http://www.windowscentral.com/how-use-resilient-file-system-refs-windows-10"><i>"Although, you may be able to select Simple (no resiliency) or Parity, both of these options will fail the process"</i></a>.
<a id="note9">[9]</a> - Storage Spaces can be used with ReFS, which supports integrity checksums, but not in Parity mode: <a href="http://www.windowscentral.com/how-use-resilient-file-system-refs-windows-10"><i>"Although you may select Simple (no resiliency) or Parity, both options will fail the process."</i></a>
</p>
<p>
<a id="note10">[10]</a> - Btrfs has external GUIs, like <a href="http://rockstor.com/">Rockstor</a>.
<a id="note10">[10]</a> - Btrfs supports external GUIs, such as <a href="http://rockstor.com/">Rockstor</a>.
</p>
<p>
<a id="note11">[11]</a> - Btrfs supports RAID5 from 2013, but recently some serious issues in the
RAID5/6 support were <a href="https://www.phoronix.com/scan.php?page=news_item&px=Btrfs-RAID-56-Is-Bad">found</a>.
At present they <a href="https://www.phoronix.com/scan.php?page=news_item&px=Btrfs-RAID56-Not-All-Clear">aren't fixed yet</a>.
The official wiki says (2025):
<a href="https://btrfs.readthedocs.io/en/latest/btrfs-man5.html#raid56-status-and-recommended-practices"><i>"There are some implementation and design deficiencies that make it unreliable for some corner cases and the feature should not be used in production, only for evaluation or testing"</i></a>.
<a id="note11">[11]</a> - Btrfs has supported RAID5 since 2013, but serious issues in RAID5/6 support were <a href="https://www.phoronix.com/scan.php?page=news_item&px=Btrfs-RAID-56-Is-Bad">identified</a>. As of 2025, these <a href="https://www.phoronix.com/scan.php?page=news_item&px=Btrfs-RAID56-Not-All-Clear">remain unresolved</a>. The official wiki states: <a href="https://btrfs.readthedocs.io/en/latest/btrfs-man5.html#raid56-status-and-recommended-practices"><i>"There are implementation and design deficiencies that make it unreliable for some corner cases. The feature should not be used in production, only for evaluation or testing."</i></a>
</p>
<p>
If you think that some information reported is incorrect or simply unfair, please report it in the <a href="https://sourceforge.net/p/snapraid/discussion/1677233/">Forum</a>.
If you believe any information is incorrect or unfair, please report it in the <a href="https://sourceforge.net/p/snapraid/discussion/1677233/">Forum</a>.
</p>
@ -294,4 +255,3 @@ If you think that some information reported is incorrect or simply unfair, pleas
</html>