RAID LEVELS DEFINED
RAID (Redundant Array of Independent Disks) is a way of storing data on two or more drives in combination for Failure Tolerance, Redundancy and/or Performance gains. By placing data on multiple disks configured as a single volume I/O (input/output) operations can overlap improving performance, or by mirroring multiple disks thus increasing the mean time between failures providing some Failure Tolerance.
Important: When using any RAID be it a RAID 1 (Mirrored) or a Hardware based RAID 5 your data is still at risk. There are innumerable ways to lose or corrupt data and hardware
failure is only one of them. You MUST have at least one complete backup of your data, preferably two, one being off-site. (Computers get stolen, offices burn, mistakes occur and there's always acts of God to consider)
You can replace your computer but you can't replace the photos you've taken over the last 10 years or the HD Video your crew spent the last month editing. KEEP YOUR DATA BACKED UP! This is as important as capturing the
data in the first place. Make it a priority.
RAIDs are not for everyone, in fact most people should avoid them and stick with a backed-up JBOD system managed by software. In some instances using a RAID actually puts your data more at risk. (Striped RAID arrays lose their data if one member fails. When you combine multiple drives in an array you increase the chance of a single drive failing by the number of drives included, i.e., a four drive RAID 0 array would have four times the chance of a drive failing) Sometimes this can't be helped such as Video Editing where the only way to achieve the needed data throughput speed is striping the disks together to gain their combined speeds and volume. In this case you'll want to have a good Backup Routine in place that backs up to an array of equal size every few hours or so. This way if you fail an array you can keep working by using the backup copy as your main data storage while repairs are made to your first array. This is why we recommend that you have a second backup. You never want to work on your only viable copy, it's like walking a tightrope without a net.
NOTE: Windows 32 bit Operating Systems can only access volume sizes of 2TB or smaller. This means that if you have a RAID Controller that creates a RAID Volume of 2+TB Disk Management will not be able to initialize it. There are a couple of workarounds for this. 3ware has an Auto Carving feature that will break a large RAID Volume into 2TB chunks so the Windows OS can manage it. RAIDs build in Disk Management can be of any size. Windows 64 bit based OS's don't have this issue.
- RAID 0 - Striped Disk Array - Without Fault Tolerance: Combines the speed and volume of multiple drives.
This technique has striping but no redundancy of data. It offers the best performance but no Fault Tolerance. By far the most common means of attaining the maximum data throughput speeds needed for HD Video editing.
It combines the individual speed and volume of each member to produce screaming fast reads and writes. If the sizes of the drive members are different, the smallest member will limit the overall size of the RAID Group.
If one drive fails or is removed then all data in the array is lost.
- RAID 1 - Mirroring: Creates drives identical to one another. This type consists of at least two drives that duplicate the storage of data. There is no striping involved. Read performance is improved since either disk can be read at the same time but write performance is the same as a single disk. This RAID is mostly useful for server applications where 100% uptime is required. If one disk fails the other has a real-time version of the data. The problem with RAID 1 is that if the data is accidentally erased or if the data becomes corrupted it's gone from both drives. It doesn't protect you from software problems, file corruptions or human error. The only protection it provides is for a hard drive failure. For this reason Mirroring is not a useful Backup Routine. If you're going to maintain a storage system it might as well protect your data from every eventuality.
- RAID 0+1 - A Mirror of Stripes: A pair of identical RAID 0s. Two RAID 0 Stripes are created (each with at least two drives), then a RAID 1 Mirror is created over them. This provides better performance than a RAID 1 but at a higher cost. The same issues apply here as far as the array only being protected from a drive failure. It provides data redundancy with performance, but does not qualify as a backup.
- RAID 3 - Striped Array with a Dedicated Parity Drive. This type of array moves data in large chunks instead of at the bit level like a RAID 5. Parity data is stored on a single drive which allows a rebuild after a drive failure while maintaining the stored data. This type of a RAID is most commonly used for large single files such as video or audio. Check out our SCSI based HUGE Systems MediaVault 320R systems.
- RAID 5 - Distributed Parity: Sometimes called Rotating Parity. Parity refers to the technique of checking whether data has been lost or written over at the bit level when it's moved from one place in storage to another. Should you fail a drive it can be removed and replaced and the data on the missing drive recreated on the fly by using the parity data as a reference. One of the drawbacks is that it has a lot of calculations taking place making it a very complicated system with a lot of overhead. RAID 5s require intense CPU time to manage the parity information. For this reason it's usually best to use a Hardware based RAID Controller that has a dedicated onboard CPU and RAM specifically for this task to avoid taxing system resources. RAID 5 due to it's complexity does not produce the super-fast throughputs that a RAID 0 would but it does have some Failure Tolerance built in. Due to the parity data being stored some storage volume is lost. For the Mac enthusiast we have the 3ware Sidecar hardware based System and the CalDigit HDPro that supports RAID 5 and for the PC's we currently have the full line of 3ware Hardware Cards for PCI-X and PCIe as well as the CalDigit HDPro.
Here's how it works (For the Geeks): An additional binary digit (the parity bit) is added to a group of bits that are moved together. This bit is used only for the purpose of identifying whether the bits being moved arrived successfully. Before the bits are sent they are counted and if the total number of data bits is even, the parity bit is set to one (1) so that the total number of bits transmitted will form an odd number. If the total number of data bits is already an odd number, the parity bit remains or is set to 0. At the receiving end, each group of incoming bits is checked to see if the group totals to an odd number. By doing so the parity data stored amongst the drives allows the controller to calculate what should be on a missing drive, allowing the ability to repair a failed array.
- RAID 5+0 -Sriped RAID 5. Two or more RAID 5 volumes are striped together to form a RAID 0. This provides the combined storage volume and speeds with maintained Failure Tolerance. A drive can be failed in either RAID 5 set and the data will be retained allowing the array to be rebuilt.
- RAID 6 - Distributed Parity: Same thing as a RAID 5 except it uses two sets of revolving parity and can therefor fail two drives. Slightly safer than a RAID 5 array at slightly higher cost. Check out the 3ware 9650 series PCI Express cards.
- JBOD - Just a Bunch Of Disks: Sometimes geek-speak just makes sense. This is where all of your mounted hard drives are independent from one another. Each drive is it's own entity. For photographers this is usually the best system. Drives can be managed by the job, location, customer or just chronologically. Each drive can be named for easy reference and should have at least one backup. This is by far the simplest and easiest to manage system of storage.
NOTE: Zero out Data - During testing we have found many benefits for erasing new drives and writing a pass of zeros to them before formatting them into an array or storing data on them. This will allow the new drives to test and repair or map as needed every bit on the platter. It makes the drives faster and a little more stable when used as a RAID. This is accomplished in a Mac by using the "ERASE" function in Disk Utility and choosing the "ZERO OUT DATA" option found under "SECURITY OPTIONS". There are many utilities written for the PC platform that perform the same function such as Western Digitals free "
Data Lifeguard Diagnostics" software. Make sure you know what drive you're working with, it's all too easy to erase the wrong drive.
NOTE: Drives must never be filled to their capacity. If drives become too full you stand the risk of overwriting the index or causing data corruption. Keep your data volume under 70% of the drive capacity for photos and OSs', stop at 80% for video.
