What is RAID 5?

Want to combine hard disks to form a RAID (Redundant Array of In­de­pend­ent Disks)? Then you’ve got a few different options. One of the most common levels is RAID 5. This com­bin­a­tion of three or more data carriers is char­ac­ter­ised by a good balance between high per­form­ance and security, and costs. So, what makes RAID 5 a good level in terms of its price-per­form­ance, and how does it work?

RAID 5 is a com­bin­a­tion of three or more hard disks that function as a single logical drive and thereby out­per­form in­di­vidu­al data carriers in terms of re­li­ab­il­ity and read speed. But to offer these ad­vant­ages, a RAID 5 system relies on two active methods. On the one hand, the network must dis­trib­ute files evenly over all clamped disks, which is also known as “striping”. On the other hand, a RAID 5 cal­cu­lates parity in­form­a­tion for all stored data, which is also dis­trib­uted to the various storage media. Using a XOR link, the storage system then enables the re­con­struc­tion of lost or damaged data blocks.

The total capacity a RAID 5 system offers for user data can be cal­cu­lated using the following formula:

For three storage media with 1 terabyte (TB) each, this results in a capacity of 2 terabytes. The remaining 1 terabyte is blocked for parity in­form­a­tion. Re­gard­less of the total storage size, a RAID 5 is only at risk if at least two hard drives fail sim­ul­tan­eously. That’s why, typically, an odd number of data carriers, i.e., three, five, seven, etc., is combined.

RAID 5 is char­ac­ter­ised by a good price-per­form­ance ratio due to its efficient gen­er­a­tion of re­dund­ancy. In contrast to other systems, the files are not saved in multiple versions, but only re­dund­antly thanks to the parity blocks. Compared to use of in­di­vidu­al drives, the storage capacity decreases, but RAID 5 networks retain a com­par­at­ively large chunk of the original capacity. On the other hand, RAID 5 is a cost-effective solution to boost read speed. The striping of the data allows parallel access to several parts of a related data block, so that inquiring devices can complete the read process much faster.

Another strength of RAID 5 is increased re­li­ab­il­ity. If a hard disk fails because it is defective, or if the data on a disk is lost for other reasons, op­er­a­tions can still be main­tained. Since the remaining storage media are used more heavily during the recovery process, the risk of failure increases sig­ni­fic­antly during this period.

However, RAID 5 is not without its dis­ad­vant­ages. Each write process to the hard disk cluster is connected with an ad­di­tion­al read step in order to check and re­cal­cu­late the parity in­form­a­tion that is available. Another step is needed to dis­trib­ute the parity data for the newly stored user data on the disks. Compared to the in­di­vidu­al drives and other RAID levels such as RAID 0, the write speed of the data carriers in a RAID 5 system is much lower.

RAID level 5 offers a good com­prom­ise between cost and per­form­ance op­tim­isa­tion compared to separate hard drives. The network also scores high for re­li­ab­il­ity. However, the reduced write rate means the storage solution is of little interest for databases with multiple larger files. In other words: RAID 5 is best used for ap­plic­a­tions that access multiple small file blocks. Typical ap­plic­a­tion scenarios of RAID 5 therefore include servers for mi­cro­trans­ac­tions and database servers, which can contain a large number of entries, but are clearly limited in terms of maximum file size.

Data re­dund­ancy with parity, which is what makes RAID 5 so efficient, isn’t unique to the level. RAID 6 also relies on the principle and even dis­trib­utes the parity in­form­a­tion twice to the in­teg­rated data carriers. Levels RAID 1 and RAID 10 take a different approach. Here, data is fully mirrored, which means that these systems offer one hundred percent re­dund­ancy. RAID 0 only increases the speed with striping and offers no re­dund­ancy at all.

A detailed com­par­is­on of the various systems can be found in our com­pre­hens­ive RAID level com­par­is­on.