The fact is there is no RIGHT option for everyone. Choosing which RAID type to use that is best for you all depends on what you’re trying to accomplish. Before choosing a configuration, you need to think seriously about what’s really important to you and your server. Is it speed, performance, reliability, up time, or something else. How good is your backup strategy? How critical is your data? These are all factors that will help you determine which is the right choice for you.
This article considers each RAID type and it’s pros/cons. Hopefully it will help you to make an informed decision when initially configuring your RAID array.
Different RAID Types to Choose
RAID 0 – This is a basic stripe array with no redundancy.
Pros – Fast Speed & Uses full storage capacity of all disks
Cons – Any one drive fails the data is gone, no redundancy
RAID 1 – Simple mirror array (generally of just two disks)
Pros – Can sustain one drive failure and continue to operate.
Cons – No speed advantage over single drive, expensive (must buy double the number of drives)
RAID 3 & 4– (Obsolete) – If you have a modern controller it doesn’t even support these. They are all similar to 5, however used a single fixed parity drive. The problem is that a fixed parity drive is extremely prone to premature failure caused by the extra read/writes during operation.
RAID 5 – Stripe Array with a single floating parity
Pros – Can sustain failure of any one drive, speed advantage similar to 0, but only “wastes” the size of a single disk regardless of the number of disks in the array.
Cons – Can’t sustain more than one simultaneous failure, not quite as fast as RAID 0, requires one extra disk than storage capacity.
RAID 6 – Stripe array with a double floating parity
Pros – Can sustain two simultaneous failures of any two drives, somewhat faster than RAID 1 or a single drive with a good controller, only “wastes ” the capacity of two drives no matter how many drives are in the array.
Cons – Not quite as fast as RAID 5, “wastes” the storage space of two drives. Costly to rebuild for data recovery if it fails.
RAID 10 or RAID 1+0 – Combination of RAID 1 & 0. Essentially two RAID 0 arrays mirroring each other.
Pros – High speed & can sustain multiple failures as long as they don’t happen to be mirrors of each other.
Cons – Very Expensive, 2x number of hard disks for storage space needed (in large arrays – very wasteful), can sustain multiple drive failures but only if it isn’t two that contain the same data (technically more risky than RAID 6 to your data though some guys might try to argue against that).
RAID 50 or RAID 5+0 – Combination of RAID 5 and RAID 0. Basically just two RAID 5 arrays together in a RAID 0 configuration.
Pros – Very high speed possible with a good controller. Able to sustain a single drive failure, two if they are from separate RAID 5 sets.
Cons – Can’t sustain two drive failures from the same RAID 5 set. Costly to rebuild if the RAID array fails.
RAID 5 with hot spare vs. RAID 6 Which RAID type to Choose?
From a data protection & recovery perspective the clear choice is RAID 6. You’d be shocked how often a second drive fails while rebuilding the hot spare, leaving the array broken. Oftentimes there is a second drive in the array that was already struggling with bad sectors but hadn’t been detected by the controller, until a rebuild is attempted and it completely fails.
With RAID 6 you can afford two simultaneous losses and still rebuild. There are some performance disadvantages of RAID 6 especially if you are running on older hardware, but if you have a good controller card, you won’t notice much impact going to RAID 6 verses 5 with a hot spare.
Here in our data recovery lab, where we constantly rebuild RAID arrays, we use RAID 6 for our data. In fact, we almost never get in RAID 6 arrays for recovery. RAID 5 w/ hot spare… we get those in all the time. So in my professional opinion, go with RAID 6 so long as you have a good controller.