What is the other alternative technique of RAID?

Question

I've been searching about this topic but I keep getting RAID only. I was told that there is another technique for data recovery and redundancy that replaces RAID. Kindly, can someone enlighten me.

Thank you :)

Solution 1

Having a RAID array is, at least from what I have seen in my professional career, the best option for redundancy - I also want to mention that having a RAID is not the same as having a backup - Even with a RAID system, having a backup of the RAID is definitely great practice.

As for alternatives, you have options such as Software Raiding (ones that are build directly into the OS or using external hardware/software to create a NAS), using "rsync" along with an application like "automator" (if you're on a MAC) will allow you to backup any directory to any locations, as long as you can provide the script for it.

With the way storage and data has been moving, everything is starting to take backups to cloud, and with internet speeds starting to move faster and faster (such as 10gb ports), RAID'ing will soon be the "older" way of redundancy.

And to end with it, RAID'ing is not the same as a backup - Performance increases and redundancy yes, but not a "replacement" for a backup

Solution 2

A technique somewhat similar to RAID but slightly different is the usage of ditto blocks with ZFS.

Unlike RAID which leverage the fact multiple storage disk are used to store redundant data that allows picking/reconstructing the originally stored data should one of the disk partially or totally fails, ditto blocks are used to spread multiple copies of the stored data on the very same logical volume. As a logical volume can be based on a single disk, the technique is similar to RAID1, a.k.a. mirroring.

This only protects against partial disk failure (i.e. a limited number of sectors are unusable) as obviously a whole disk failure is unrecoverable unlike with real RAID.

An interesting functionality of ditto blocks is their flexibility. Ditto block presence can be (and actually are always) a function of how critical the information stored is. For example, top level metadata can be stored three times, regular metadata two times, regular data either once, twice, or more, depending on the dataset.

Note that ditto blocks can be used together with RAID.

Comments

[ganesh]

"I am told there is something else" is rather vague.Answer can vary from ZFS to backups to having multiple disks and manually copying stuff.

[Frank Thomas]

There are many types of RAID and other multi-disk arrays (JBOD, BIG/SPAN, etc) but no family of technologies has supplanted RAID as the dominant approach to disk redundancy at the hardware level.

[David Schwartz]

RAID is not a technique for data recovery. It's just a way to continue operating even in the event of particular types of hardware failures.