• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

What Usage Cases Benefit More From Hardware RAID VS Soft RAID

M

MySongRanHills

Limp Gawd
Joined
May 27, 2011
Messages
237
I've administered various RAID levels (0,1,5,6,60) for years and various implementations (ZFS, mdam, FlexRAID/SnapRAID, various hardware RAID cards). While recently migrating a hardware RAID-5 to RAID-6 on a Dell server with a PERC H710p RAID card @ my office , I started thinking about what other points of failure we had and what I could do to mitigate them.

So now that everything is on RAID-6 with all new HDDs I figure next most likely failure would be the RAID card itself. My understanding is I would need to get an identical PERC H710p to be able to import the config. I would never have this problem on ZFS or other software RAIDs. If at home my 18 disk ZFS system had the motherboard /HBA/SAS expander all die at once, I'd just find a spare computer with enough SATA ports and import the pool.

So this begs the question (to me atleast) why ever use hardware RAID? I'm sure there are some cases where it is better ,but I can't think of any at the moment.
 
What additional features? As for performance modern CPUs don't break a sweat during parity calculations?
 
To me battery backed cache is the only reason on a non windows machine. On a windows machine the performance of software write back cache is extremely slow (compared to linux) in every implementation that I have seen.
 
MySongRanHills said:
I've administered various RAID levels (0,1,5,6,60) for years and various implementations (ZFS, mdam, FlexRAID/SnapRAID, various hardware RAID cards). While recently migrating a hardware RAID-5 to RAID-6 on a Dell server with a PERC H710p RAID card @ my office , I started thinking about what other points of failure we had and what I could do to mitigate them.

So now that everything is on RAID-6 with all new HDDs I figure next most likely failure would be the RAID card itself. My understanding is I would need to get an identical PERC H710p to be able to import the config. I would never have this problem on ZFS or other software RAIDs. If at home my 18 disk ZFS system had the motherboard /HBA/SAS expander all die at once, I'd just find a spare computer with enough SATA ports and import the pool.

So this begs the question (to me atleast) why ever use hardware RAID? I'm sure there are some cases where it is better ,but I can't think of any at the moment.
Click to expand...

The only real reason, IMHO is simplicity... you don't need to manage/configure ZFS and pretty much any admin can setup hardware raid and go. Hardware raid requires the batteries to get replaced every 1-2 years that's $$ and time that adds up.
 
drescherjm said:
To me battery backed cache is the only reason on a non windows machine. On a windows machine the performance of software write back cache is extremely slow (compared to linux) in every implementation that I have seen.
Click to expand...

This makes sense. I can see why people want battery backed write cache. Personally give me a good UPS instead.
 
drescherjm said:
To me battery backed cache is the only reason on a non windows machine. On a windows machine the performance of software write back cache is extremely slow (compared to linux) in every implementation that I have seen.
Click to expand...

You can add a L2ARC device (SSD, NVME) to ZFS for cache drive that persists after power-loss, you can use high-end SLOG devices for ZFS that have > capacity than RAM on hardware RAID and also persist after power less. (NVME, ZuesRAM). Your entire server in a business environment likely is already battery backed, and may have redundant power sources, and you then have 1 less 'battery' to replace too.

With that said I don't use ZFS for simple raid1 setups.
 
The basic question is not hardwareraid vs softwareraid but any type of raid with a
conventional filesystem vs softwareraid with a CopyOnWrite filesystem like btrfs, ReFS or ZFS.

The reason is mainly the write hole problem and the raid behaviour on problems during a write.
As disks in a raid must be updated sequentially, a crash during a write can result in two different
datastates in a mirror or a raid stripe error. This can mean a corrupt raid state and/or a corrupt
filesystem. The worst thing, with a conventional filesystem is that you cannot detect nor repair
these problems.

This is why you should use hardwareraid with a cache and a BBU (that is needed to hold data
until next reboot) and a UPS for security on these filesystems.
On a crash the otherwise last lost write (the last raid stripe or metadata update) can be done
on next bootup. This does not mean that the file that was written on a crash is not corrupt, but
it means that there is at least a chance that the raid and the filesystem structure remain intact.

see http://www.raid-recovery-guide.com/raid5-write-hole.aspx about the problem.

A raid with a Copy on Write filesystems behaves different.
If you write a datablock, its is written completely or not what eleminates the problem with
a partly written raid write action. It additionally eleminates the problem if you have different
data ex on a mirror (what can also happen due a silent data error problem) as it adds checksums
to all data what allows a repair of an inconsistent raid 1 or raid-Z or data errors on disk in general.

So the basic usage difference when you care about security.
- use hardware raid with BBU and cache with old filesystems like ext4, hfs+ or ntfs
- use softwareraid with btrfs, ReFs or ZFS (with ReFS you must enable data checksumming)
This gives you a much higher level of datasecurity beside a better performance and expandability.
 
_Gea said:
The basic question is not hardwareraid vs softwareraid but any type of raid with a
conventional filesystem vs softwareraid with a CopyOnWrite filesystem like btrfs, ReFS or ZFS.


So the basic usage difference when you care about security.
- use hardware raid with BBU and cache with old filesystems like ext4, hfs+ or ntfs
- use softwareraid with btrfs, ReFs or ZFS (with ReFS you must enable data checksumming)
This gives you a much higher level of datasecurity beside a better performance and expandability.
Click to expand...
Wow, great explanation. I learned something today, thanks for the info.
 
Don't go with FlexRAID. When you buy the license it's locked to whatever hardware you initially install it on. Once you begin to upgrade/replace components you'll eventually be at a point where not enough of the "old hardware" is the same and thus you'll be forced to buy a new license from them. Avoid them.
 
You must log in or register to reply here.
Back
Top