RAID 5 is the most popular RAID configuration for both enterprise NAS devices and business servers, RAID 5 provides better fault tolerance and performance.
Whether you are using RAID for fault tolerance or for performance or even both, it is important to understand that there are different levels of RAID.
The one you choose depends on what you are using RAID for.
Also, another determinant factor is the type of RAID you are using – software or hardware RAID.
Since hardware-based RAID supports more levels than software RAID, the type of controller you have can also determine your choice.
The type of disks you can use in an array (SSD, SATA or SAS) can be dictated by several controllers which also support different RAID levels.
If you are looking to have a RAID level this is suitable for DB – heavily read oriented, providing good performance and redundancy and cost effective at the same time, then you need RAID 5.
What is RAID 5?
Without any doubt, RAID 5 is the most popular RAID configuration for both enterprise NAS devices and business servers.
Unlike the mirrored disk sets, the parity disk protection of RAID 5 provides better fault tolerance and performance.
You can easily stripe additional data used for recovery across 3 or more disks, i.e. data and parity with this RAID level.
Unlike some levels of RAID, the stress of a dedicated parity disk is evenly distributed among all RAID members on RAID 5.
As all members of the RAID participate in the serving of the write requests, write performance tends to increase thereby eliminating every delay in progress.
However, it is important to note that since parity must still be written, it will still not be as efficient as a non-RAID setup.
From this distributed data and parity block, data can be automatically and seamlessly recreated once a disk starts to fail or gets an error.
In essence, the system remains fully operational even when one physical disk fails or crashes and until the failed drive is successfully replaced, the system will still be in a functioning state. By calculating and storing parity, fault tolerance can be achieved.
Additionally, it is good to know that many NAS and server drives can be “hot-swappable” with RAID 5.
By implication, a failed in the array can be efficiently swapped with a new drive without causing any form of interruption for users accessing the server or NAS or without shutting down the server or NAS.
Drives don’t last forever, they will at one point in time or the other fail.
However, this RAID level is proven to be a reliable solution for fault tolerance. As failing disks are being replaced, the data can be effectively rebuilt to new disks.
In RAID 5, parity information is well-distributed among the drives. While consisting of block-level striping with distributed parity, this RAID level demands that all drives but one be mounted for operation.
Subsequent reads can be calculated from the distributed parity upon failure of a single drive such that no data gets lost. However, nothing less than 3 disks is required to make this work.
Unfortunately, servers that perform a lot of write operations may experience underperformance with this RAID level.
There could be a noticeable lag with RAID 5 on a server with a database accessible by many employees during working days.
RAID 1 vs RAID 5 (Compare and Contrast)
In a bid to provide redundancy and fault tolerance, the same data can be stored on 2 or more physical disks.
This simple mirror configuration is known as “RAID 1.” The data is simultaneously written to both disks when a write is carried out on the mirrored pair of drives.
This is why RAID 1 is commonly referred to as a mirrored pair of disk drives.
Here is a detailed representation of RAID 1 and RAID 5
Mirroring, Redundancy and Fault Tolerance
While RAID 1 offers full redundancy and fault tolerance through a simple mirror configuration, RAID 5 provides no mirroring or redundancy.
Although RAID 5 also tends to provide fault tolerance, however, data is striped with parity across several disks.
In RAID 1, data is fully stored on each disk, so no stripping is allowed. However, in the RAID 5 setup, data can be split or striped evenly across multiple disks.
Additionally, if one of the drives fails, data can be recovered because parity information has also been stored.
Although RAID 1 is generally known to perform slowly, however, if multiplexing is utilized by the RAID to read data from disks, the same read performance as RAID 0 could be offered.
Due to its ability to distribute data across multiple physical disks, RAID 5 tends to offer faster reads.
Nevertheless, since parity information must be calculated, writes have become a little slower.
Minimum number of disks required
While RAID 1 requires a minimum number of 2 physical disks, RAID 5 requires 3.
In RAID 1 data loss is generally unacceptable. RAID 5 offers a good balance of adequate security, failure resistance, decent performance, and efficient storage.
File and application servers with a limited number of data drives work better on a RAID 5 setup.
In RAID 5, parity information is evenly distributed among the available disks. This info helps to recover data that was once stored on a failed disk. But in RAID 1, no parity disk is used.
Although writes are a little slower with RAID 1, nevertheless, it still offers great performance and efficiency.
Even while a failed drive is in the process of being rebuilt, data can still be accessed on a RAID 5 setup (though at a slower rate).
This RAID level encourages inexpensive redundancy and fault tolerance, as well as fast reads.
During recovery, data cannot be accessed because it is a process that requires powering down the RAID.
Also, since 2 copies of all data are stored, storage capacity is effectively cut half.
As a result of parity calculations involved in rebuilding a replacement drive and restoring data, recovery from failure is often very slow.
During the recovery process, it is possible to read from the RAID, however, this can also be very slow.
How to use RAID 5 Calculator
Extra disk space is required to store parity data on a RAID 5 setup.
This helps in rebuilding the lost data onto a new drive which might be due to a drive failure. Although parity data is evenly spread among all disk, however, for parity information, one additional drive per RAID group is required.
Bear in mind that the parity overhead will take up one drive’s worth of capacity if the available 8 drives were taken thereby leaving a total of 1,022 GB – 7 (146 GB drives) – for data.
In preparation for the creation of LUNs, all the drives will be formatted when a RAID group is created on a storage array.
As a result, the capacity of the RAID group will be reduced by about 5 percent to 10 percent.
In order to obtain an idea of the total amount that will be available for use, it is best you reduce the available capacity of the disk drive by about 15 percent.
This is a simple rule for RAID 5 calculation.
If you know how to calculate RAID 5, you will be able to rebuild the missing data which may have occurred as a result of a drive failure.
This will help you when adding one drive for parity. In order to calculate the RAID 5 effectively, there is need for a very efficient XOR engine.
This XOR does not have to deal with 3 bits but only needs to deal with 524288 bits. 64k (65536 * 8 = 524288 bits) is the most common stripe size employed in RAID 5.