I did write about my sever structure with a bit of explanation on Redundant Array of Independent Disks (RAID) before. But today I came across an email I sent out to one of my friends explaining how RAID works and what each type of RAID benefit each situation. Today I will post the email as it is and after the final exams, I will rewrite this article to include more specific details.
RAID(redundant array of independent disks) protection technology is developed to safe guard data in an event of a hardware failure. I found that it is easy to use online info than to type all the explanations. You can read about RAID and watch the following movies to learn the technical information:
RAID 0(not really RAID) : http://www.youtube.com/watch?v=Uwie0rJSYiE
RAID 1: http://www.youtube.com/watch?v=-PlS4seeexA&feature=related
RAID 5: http://www.youtube.com/watch?v=LTq4pGZtzho
http://www.codinghorror.com/blog/2009/05/beyond-raid.html
There are few types of RAID and each one has its unique qualities and drawbacks. Most companies that provide server services utilize more than on type of RAID to provide maximum performance and data redundancy. The types of RAIDs as follows:
– RAID 0: This is not true RAID since it won’t provide drive failure protection at all. In this type, all hard drives were taken as a one single entity to provide maximum data storage capacity. The formula for RAID 0 is: Disk 1 + Disk 2 + Disk 3 + n = total capacity, where n can be any number of disk. But, the maximum capacity is limited by the multiples of the lowest capacity drive. For example, if I have 50 GB + 100 GB + 500 GB drives the total would be 50 GB X 3 = 150 GB, not 650 GB. It is advised to use same capacity disks for for a single RAID array to prevent wasted space and reduction in performance. The biggest drawback would be if one drive fails, all data will be lost regardless of health of the other drives. The data is stripped and stored on all devices where one bit for a single file will be saved in one drive and the other bit will be in another. So, when a drive fails, there will be missing bits for the file causing data recovery impossible unless you recover the fail drive successfully.
– RAID 1: This is the maximum data protection type where data is copied on two hard drives as mirror image of each other. If a drive fails, the data can be recovered by the second drive. In fact you can continue to use data while a drive has already failed. The only thing you need to do in an event of a hard disk failure is to replace the failed unit with a new one with exact specifications. The disadvantage however is you have to use exact same size of drives and you lose the data storage space by exactly half. The formula for the RAID 1 is: (Disk 1 + Disk 2)/2 = total capacity, where you lose half of the data space. For example, if you have two 500 GB hard drives, the system will only allow 500 GB of data to be stored rather than 1 TB. Also, you are limited to two drive of protection in pure RAID 1. However, there are ways to modify this with other RAID types to utilize more than two drives.
– RAID 5: This is consists of block-level striping with distributed parity. Unlike in RAID 4, parity information is distributed among the drives. It allow increase the data capacity while keeping one harddrive failure tolerance. One of the pitfalls of RAID 5 is that when a drive fail, it could take a long time for other logical drive to redistribute the data.
– RAID 1+0 (RAID 10): Requires minimum of 4 drive and often used in Linux environments.
You may use an online RAID calculator to estimate the driver space for storage in each RAID option. However, the storage capacity may change slightly depending on the RAID (hardware RAID or software RAID) is handled.