RAID 1 mirroring for backups

RAID-1 isn't a backup strategy, it's a way to survive a disk failure. Cycling through drives is also asking for trouble and for more load with RAID-1, as macOS lacks any way to optimize how much data gets copied. It's always full copies and not incremental copies, it's also using arcane to create and re-create commands and it's (by default) a manual process to perform the clones, and it's quite possible to end up aiming the copy source and the copy target at the wrong storage devices. Whoops.

RAID-1 is also a way for a lurking corruption or a deletion to propagate throughout all the backups.

There's also only as much "depth" to a RAID-1 backup scheme as you have disks in your backup pool.

Use Time Machine and a big Time Machine disk. It's built in software, it's restorable, it's automatic, and it's largely transparent, and it's something that you and other folks can help with questions and such. And it has as much depth to your backups as you have storage on your Time Machine Target disk, and it avoids duplicate files on the backups.

If you're out of internal storage on your internal storage device, configure a ThunderBolt or USB-C/USB3 external storage or external hardware RAID array, and use a separate Time Machine backup target storage configured to record the internal storage and the external RAID array. Or create and use multiple partitions on the RAID array, if you're using mid- to upper-end RAID hardware.

If you're particularly paranoid about the data, configure multiple Time Machine targets (across separate disks and/or separate RAID arrays) or separate Time Capsule devices scattered around, and disconnect and rotate the directly-connected Time Machine backups. Those'll automatically re-synch for you too, unlike the RAID-1 array members that'll require the full copies to resynchronize the contents. Or upload the source files and the finished files to a cloud provider and distribute from there. Or a combination of these.

If you really want to go with clones, maybe Carbon Copy Cloner and set that to create clones for off-site or otherwise.

But I'd not use RAID-1 as a backup strategy here. Software RAID-1? That's centrally intended to allow your system to survive the failure of storage device. Not backups.

ps: Yeah. I've used RAID-1 the way that you're considering. On systems with what I'd consider far more robust RAID-1 software support, too. For the vast number of environments, Time Machine is far superior.

Using a traditional backup to a RAID device, in my opinion, is great. I send mine to a 5-bay Drobo at home, and I have a 4-bay Drobo at work.

But for me splitting a mirror as a backup strategy means to much re-silvering the mirror when you replace the removed disk with another one that now needs to be mirrored to the existing disk.

And BulletRouge wants 2 backup devices, so one can go off-site. I assume the backup devices would rotate between being on-site and moving off-site. So a RAID device doubles the cost for backup devices. But as I said, I personally use a Drobo for my RAID backup storage.

I use a RAID mirror for one of my backups. Mirror RAIDs offer data redundancy and recoverable data even if one of the drives fails. Large and small data centers utilize RAID arrays for both storage and backup. However, removing a drive for an off-site location makes no sense whatsoever. Cloning the array to another drive that goes off-site makes more sense and provides another level of redundancy.

RAID Basics

For basic definitions and discussion of what a RAID is and the different types of RAIDs see RAIDs. Additional discussions plus advantages and disadvantages of RAIDs and different RAID arrays see:

RAID Tutorial;

RAID Array and Server: Hardware and Service Comparison.

Hardware or Software RAID?

RAID Hardware Vs RAID Software - What is your best option?

RAID is a method of combining multiple disk drives into a single entity in order to improve the overall performance and reliability of your system. The different options for combining the disks are referred to as RAID levels. There are several different levels of RAID available depending on the needs of your system. One of the options available to you is whether you should use a Hardware RAID solution or a Software RAID solution.

RAID Hardware is always a disk controller to which you can cable up the disk drives. RAID Software is a set of kernel modules coupled together with management utilities that implement RAID in Software and require no additional hardware.

Pros and con

Software RAID is more flexible than Hardware RAID. Software RAID is also considerably less expensive. On the other hand, a Software RAID system requires more CPU cycles and power to run well than a comparable Hardware RAID System. Also, because Software RAID operates on a partition by partition basis where a number of individual disk partitions are grouped together as opposed to Hardware RAID systems which generally group together entire disk drives, Software RAID tends to be slightly more complicated to run. This is because it has more available configurations and options. An added benefit to the slightly more expensive Hardware RAID solution is that many Hardware RAID systems incorporate features that are specialized for optimizing the performance of your system.

For more detailed information on the differences between Software RAID and Hardware RAID, you may want to read: Hardware RAID vs. Software RAID: Which Implementation is Best for my Application?

One cannot have enough backups so your backup strategy needs to include more backups. The array is only one. Two arrays even better. Adding a Time Machine backup even better.

If you encrypt your drives, including the Time Machine drive, then I start to be a little nervous. Forgetting an encryption key means all is lost. Time Machine in my experience is not 100% reliable. I prefer clones or multiple, similar backups using file by file copies. Now, I am limited to the loss probabilities of drive failure and my backup software. No different than using Time Machine exclusively. But I don't have to rely on somewhat convoluted ways of restoring a few files that require using a separate app. Time Machine also demands much larger backup drives so you don't have to erase the backups and start over or replace them with new backup drives.

MrHoffman, above, knows his stuff but I have to disagree with him here.

RAID is not a substitute for a backup. The singular benefit that RAID provides is to reduce the exposure to an isolated failures of storage devices, and sometimes not even then. RAID does nothing for volume corruptions. Nothing for accidental or malicious deletions. Nothing for failed upgrades. Malware. Etc,

Once you start cycling drives in and out of the RAID-1 set, there are negligible differences between cycling volumes through a RAID set and cycling scratch volumes for full copies, and that then becomes a question of how often and how much effort is involved and how automatable and how much exposure to errors each approach entails. As well as the usual sorts of problems with dropped disks and forgotten backups.

If you're transferring around volumes, you probably want those encrypted. Otherwise whatever data is resident on the device can end up getting accessed when the device is stolen or is retired. Time Machine can encrypt the volumes as part of its normal operations. macOS on current releases with current installs also usually encrypts its boot volume.

Time Machine is also baked in and you're not going to have to write your own tooling. Carbon Copy Cloner or other tools also include the ability to automate the backups. Manual backups? Not so good. Because backups that aren't done aren't useable for recovery...

Yes, I've had corruptions happen with Time Machine. I've had both software and hardware RAID volumes fail and corrupt, too.

Time Machine isn't perfectly reliable, and nothing is of course, but it's also not a huge matter to both run multiple parallel Time Machine volumes, and to cycle through those volumes if necessary hauling some off-site, or you can back up the Time Machine volumes (not many do that, but Time Capsule and other tools can support that) and in the worst case you start over with your Time Machine backups.

BulletRouge wrote:

I was more wondering about was if (and how) one could be doing mirroring between two disks, take one disk out of the array (wondering if that drive can be used as a stand-alone copy of the mirror, and put another disk in to continue mirroring.

You'll likely write your own bash script to do that, as well as tasks such as scheduling and user prompting if that's necessary. Maybe an AppleScript that invokes bash, or a bash script that invokes AppleScript for prompting. The basic command for managing disks and particularly RAID volumes is the diskutil utility.

Find your disks:

sudo diskutil list

Find which one is a RAID volume and fetch the UUID of the RAIDset:

sudo diskutil appleraid list

Add your new disk to the RAIDset:

sudo diskutil appleraid add member disk4 {uuid-of-the-raidset}

Remove a volume from the RAIDset:

sudo diskutil appleraid remove {uuid-of-the-failed-device} {uuid-of-the-raidset}

Create a RAIDset:

sudo diskutil appleraid create stripe {arrayname} {volumeformat} {list of disks to add}

Once you have the volumes mounted and built and unmounted, run some tests to see what's accessible and not accessible with the volumes extracted from the RAIDset, and test the recovery path.

I'd also expect to shut down and reboot around these operations, as active files might not be captured correctly from live copies — all the RAID members will be consistent, but — depending on how a multi-part file update was progressing when the RAIDset was split — the contents might be inconsistent. Probably not a big deal for only-occasionally-modified movie files, but can be a real hassle for the more active data files.

Encryption...

Rebuild...

The equivalent to diskutil appleraid for Time Machine is the tmutil utility. Though for most tasks and cases, the Time Machine GUI is entirely capable, including setting up multiple targets, and the GUI allows the user to handle file and directory restoration, among other tasks.

If you are going to use external storage, then I would suggest you get a dedicated RAID box, either direct attached (Drobo.com) or Network Attached Storage (Synology.com) vs rolling your own.

HOWEVER, as a backup solution I agree with MrHoffman that RAID is not backup.

Also if you split the RAID and then put in a new device, you are going to be re-silvering the added drive from scratch every time you swap a drive.

Where as, if you use a backup utility, such as Carbon Copy Cloner or SuperDuper, you can do an incremental update where you will ONLY need to copy files which have changed since the last time you backed up to that specific drive. That is a lot less reading and writing vs re-silvering the disk you add into the RAID.

BulletRouge wrote:

Thanks, that's actually very helpful I can read up on the command and script something myself.

So if I remove a disk from a RAID 1 array would that drive be mountable?

Sure. It'll be mountable. I'd expect the volumes will allow a normal mount, but — worst case — the volume contents can accessed and the contents dd'd somewhere else. But given you're going to necessarily be testing the full create and device cycle-through and RAID restoration sequence and whatever other related scripts once or twice before deploying them, you'll know the behavior of the whole sequence soon enough.

Try Time Machine, too. TM will very likely involve less support, less UI work, and less ongoing management, and the TM UI and TM support is baked into upgrades and system restorations and related processing.

As I said, I disagree. Maybe you could define what you see a backup to be. Frankly, all the things you say RAIDs do nothing for would apply to any backup device. I fail to see your point. There is no argument that a RAID provides data redundancy in the event a disk fails. It is hardly any different than having two separate drives and performing the same backup to each drive such that they mirror one another. That is what a mirrored RAID is.

Two identical copies that are always identical is not backup. It's redundancy. Cycling a third copy through for purposes of backup is no different than a backup, though with the risk of corrupting the primary copy as the RAID user interfaces are not intended for that usage, and requiring the wrappers and related code to be written locally, and not integrating with what Apple has established, and the clone doesn't get you a check of the file system that's inherent in a file-based copy nor does it get you any sort of compression absent hardware compression. As I have stated, I've done this. I don't and can't recommend it. TM works better for many of these applications and deals with pruning and compression and storage far better than cycled clones, and CCC has the wrappers and related supporting code to deal with all of what a cycled clone does without requiring bespoke local scripting.

What, then, is your thinking about performing a Time Machine backup to a mirrored RAID? I still don't understand why the RAID used in this way or with other backup utilities isn't a backup. Time Machine must back up to a disk. What difference does it make if the disk is single vs. the RAID mirror? The backup has to be made on some device. You haven't convinced me.

Most of the better backup utilities provide options that pretty much duplicate what Time Machine does if that is what you want. I do not require pruning or compression when doing incremental updates of the original clone. I don't need hourly incremental backups. Hence, I never run out of disk space as would be the case with Time Machine.