Need reasonably priced NAS RAID Storage using Hardware RAID controller QNAP is all software RAID

[SynapticHorizons]

There are too many recommendations for high function NAS RAID devices using Software RAID for example the QNAP TVS-h1288X 12-bay ZFS NAS recently reviewed by Patrict. These systems all have many desirable features except the one most important feature of Data Integrity which is a Hardware RAID controller. The QNAP 2-year free Data Recovery Service offered by QNAP only applies to users who configured the NAS as a JBOD device. If the user created a RAID 0/1/10/50 etc the recovery service doesn't apply. I learned a long time ago with Dell PowerVault RAID software storage what happens when things go south with software RAID. When the hardware RAID controller goes down just pop a new one in.
So does anyone have a good suggestion for Hardware RAID NAS with 10GbE/M.2 NVME slots (for high speed caching) at least eight (8) 2.5" SSD/SAS compatible hot swap disk bays. With dual controllers that has a decibel level under 50dba?

 

[BlueFox]

You're soon going to get an earful that hardware RAID is not the end-all, be-all, of data integrity (as it isn't). It's really on its way out given how high performance storage needs all the PCIe lanes you can throw at it, far beyond what any current RAID SoC can provide. Hardware RAID can fail just as spectacularly as anything else and in fact can be more of a headache in terms of recovery as things are going to be proprietary to a given vendor, and potentially specific models at that. Speaking from experience there. As always, RAID, of any kind, is solely for uptime, and has never been a substitute for backups.

You're not really going to find a NAS how you describe it so far as I'm aware. Closest would be a SAN from one of the large commodity vendors, be it Dell, HPE, etc. Never seen any hardware RAID with M.2 SSD caching either.

 

[gea]

You asume that hardware raid is always superiour to software raid regarding data integrity. This was right 15 years ago or when you look at software raid based on "old" filesystems like ext4/ntfs or in general at filesystems without "CopyOnWrite" and checksums on data and metadate.

Modern softwareraid especially based on ZFS is by far superiour in nearly all aspects ex performance, features, data protection, crash behaviour or portability ex

Write hole problem
A hardware raid adapter updates disk by disk on writes (ex a mirror). A crash between can result in a state where data state is different on both disks. As there are no data checksums you cannot detect which data or part of the mirror is valid and which is not. A BBU protection can reduce the problem but cannot avoid. ZFS can avoid problems due CopyonWrite and detect problems based on checksums on all datablocks on any disk in a raid, "Write hole" phenomenon in RAID5, RAID6, RAID1, and other arrays.

Silent data errors (random data corruption over time)
Without data and metadata checksums there is no chance to detect these problems based on a hardware raid controller. ZFS and software raid detects all corruptions and can repair on the fly

Secure sync write
A hardware raid controller can do with a cache and bbu protection but without verification. ZFS can protect its rambased writecache with an Slog and checksum protection. As ramcache is mainboard ram this is faster and bigger on ZFS.

Portability
For hardware raid, you need a compatible controller on a failure. With software raid you can just connect the disks to any controller and import the datapool.

Expandability
With ZFS and software raid, you can scale storage to hundreds of disks and Petabytes of storage and very high data rates. A modern mainboard cpu scales much better and is much faster than a hardware raid cpu. Software raid with offers fast rambased readcaches in the area of hundreds of gigabyte ram. Softwareraid can be build from or extended by NVMe disks ex caches, special vdevs or Slogs. Hardware raid is limited and was faster only a long time ago.

Above of these controller based items, ZFS offers superiour rambased read/write caches, additional SSD/NVMe readcaches with read ahead, special vdevs as a superiour tiering alternative to improve access to metadata, dedup tables or small io, superiour compress, dedup and encryption options - can be set per filesystem, sync of multi petabyte storage systems in near realtime under heavy load based on snaps. ZFS snaps can be created without delay and initial space consumption. Even 10000 snaps are no problem regarding performance and stability. Snaps are read only and offer a perfect protection against ransomware.

ZFS and software raid can give you all the superiour goodies, hardware raid cannot (even a combination of ZFS and hardwareraid would be really bad as you would loose many features).

You may read some of my manuals like ZFS basics or principles at

napp-it // webbased ZFS NAS/SAN appliance for OmniOS, OpenIndiana and Solaris : Manual

napp-it.org - free ZFS NAS SAN Server with user editable web-gui // All In One server = virtualized ready to run ZFS-Server

napp-it.org

Most of the infos are valid for any ZFS system incl. Qnap or any ZFS system, others for native ZFS (Oracle Solaris) or the Solaris forks like OmniOS (a enterprise class storage OS), OmniOS Community Edition that allows to use an average storage server ex SuperMicro, Dell or HP.

 

[Evan]

What do you think hardware raid is ? These days its software running on a CPU that happens to be embedded on a card.
it has its place sure (like useful for ESX local data stores) but so does software raid.

think about enterprise storage arrays, they are all essentially servers running software to do raid with disks out the back.