Wyered1 :
You're making it sound like there is NO USE for striping data, and I can't figure out exactly why. Would you be opposed to a RAID5 array or a RAID10 array as well?
My take on using RAID is a little more conservative than most people's. I will only choose a RAID level if 1) There is a specific problem that needs that RAID level to address it, or 2) I want to fiddle around with benchmarking and learn things. I once bought an 8-bay rackmount RAID server just for point 2. My wife was very happy when I threw it out.
That said
RAID 0 (begin rant) is not truly RAID. The "R" in RAID stands for Redundant; RAID0 loses all of its data if one spindle is lost (end rant). It has a small improvement in seek times using mechanical disks, as one disk can be transferring data while another is seeking, but not for SSDs. I would use RAID0 for data that will not be highly fragmented, must be transferred very fast, and is expendable. For example, Photoshop scratch files or a copy of a video project that I am currently working on. Note that I wrote "a copy;" I would not keep data that I want around for a long time on a RAID 0 array.
Now that the SSD has been invented, I would always use a scratch SSD instead of RAID 0. If I could afford an SSD that large, that is. If I needed 500 GB of such space, I might RAID 0 three short-stroked drives. Short-stroking is only using a fraction of the capacity, keeping seek times low by only using the outer portion of the drive.
----
Prefix to the rest: No RAID level is a substitute for backups. All of the true RAID levels let your machine keep going without losing any data if one (or more) drives fail, but you are still vulnerable to data loss due to malware, cats, rain, or a three-year-old banging your computer against the wall (an actual example, but not from me).
I would use RAID 1 to protect volatile data where I don't want to lose even a day's worth. I would still do backups, but the RAID 1 would protect me from a disk failure. I actually have one folder on my machine that is mirrored with a software driver - it's my Email. RAID 1 is rumored to speed up reads by distributing them between the two drives.
RAID 5 or raid 0+1 would come in if I needed to build a system with a shitload of data. It lets you build up capacity from smaller drives. RAID 0+1 would give me a total usable capacity of half of my physical disk space. RAID 5 with N drives gives you the capacity of (N-1) of them.
Two things to know about RAID 5. First, it is slower than a single disk until you get to a certain number of drives. I don't know what that number is, but it is greater than three, the minimum number of drives you can put in 5. Second, RAID 5 should be done with a more expensive hardware controller. Unlike 0, 1, and 0+1, the other RAID levels require parity calculations. In the olden days, this could put an enormous strain on your CPU unless you had the more expensive kind of controller that had an onboard parity processor. These days the CPUs are fast enough, but the controller to CPU to controller loop takes a toll.
If I were building a serious storage server for making a living, I might use RAID 6. It's like 5, only there are two extra drives instead of one, and different parity calcs for each stripe are written to to separate drives. Put that in a box that has a "hot spare" and you are protected against drive failures. Of course, the RAID controller could fail, the power supply could blow up, a dozen other things could still take out your data.
EDIT: This is a personal opinion. There are some members out there who are much more competent than I am who use RAID all the time, and I may or may not agree with their reasons.
Facebook
Twitter
Instagram