Did You Know?
The term 'RAID' was coined by David Patterson, Garth A. Gibson, and Randy Katz at the University of California, Berkeley, in the year 1987. What if your computer memory system crashes, and you don't have any backup of the data stored elsewhere? Clearly, that will put you in a difficult situation; especially in a professional scenario.
The RAID (redundant array of independent disks) data storage system is designed in such a way that data recovery is possible even if a storage device fails. RAID disks (tape disks or magnetic disks) are so configured that they allow redundancy (mirroring or repetition) of data to take place.
Some RAID architectures are designed for faster read operations, some for faster write operations, while others are designed for both, faster read and write operations.
RAID 1 architecture uses the concept of disk mirroring. In it, data is copied from one disk to another at the time that it is entered, creating a replica or mirror of it. Thus, if one disk crashes, the other can still continue working, and the data is retained.
For hardware RAID 1 implementation, a minimum of two physical hard disks are needed. In the software-based RAID 1 implementation, two separate volumes of a single hard disk are utilized.
RAID 1 is designed for faster read operations as compared to write. While using any modern multi-threaded operating system (Windows or Linux), RAID 1 array gives better performance for split seeks, that is, access of same memory location by two different users. For best performance, each disk in the array needs to have a RAID 1 controller.
RAID 1 implementation chops the total storage capacity in half. Thus, even if you have two 500 GB hard drives in your system, the total storage space available to you will only be 500 GB, as the remaining half will be used for the purpose of redundancy.
- Redundancy of data is hundred percent.
- Data is not required to be rebuilt in case of disk failure.
- It requires minimum hardware resources, and can also be implemented using just two disks.
- Even though this architecture may comprise only two disks in the array, the rate at which data gets transferred will be equivalent to that of a single disk.
- This configuration is easy to understand and implement.
- This architecture shows a noticeable performance degradation in the data read operation.
- Since it requires redundant drives, maintenance cost increases.
- This implementation cuts the total storage capacity in half.
- RAID 1 is not preferable for backing up large amounts of data.
RAID 5 is the most popularly used configuration in a number of business servers and enterprise NAS devices, as it is able to provide a better performance as compared to the mirroring and fault tolerance schemes of the previous RAID levels.
This architecture requires a minimum of three physically distinct hard disks. When data is written, it is initially divided (stripped) into three, each of which gets saved on one of the drives. At the same time, parity calculation takes place, and in each disk, parity information of the data stored on any one of the other disks is saved.
Parity is a much smaller piece of information derived mathematically from a larger data set, which can be used to accurately recreate the data. Thus, if one of the disk fails, data can be recreated from the parity information, preventing a loss or failure.
One of the best features of RAID 5 is that, the hard drives are hot-swappable. This means that a failed drive from the array can be directly replaced with a new one without having to shut down the NAS or the server, allowing operation to continue. Thus, failure of a disk can be kept hidden from the end user.
RAID 5 provides good fault tolerance, because when the drives fail, data is rebuilt into the new disks that take its place. The main disadvantage of RAID 5 is the lag that is typically seen on servers where a lot of write/modify operations are performed.
- RAID 5 utilizes total storage capacity efficiently, reducing it by only one single disk which is used to store parity information.
- It has an inbuilt load balancing mechanism.
- Its data read performance is better as compared to RAID 1, making it ideal for database applications which are highly read oriented.
- It is able to sustain the failure of a single drive without any data loss or affecting the functionality at the user end.
- High fault tolerance typically requires additional disk space. RAID 5 is highly efficient in such a scenario.
- The write operation is slow, as unlike in RAID 1, parity calculation needs to take place.
- A minimum of three disk drives are needed to implement this architecture, thus, the total hardware requirement is more.
- The implementation of RAID 5 is complicated as compared to that of RAID 1.