What is RAID 10?

While cost is no longer the deciding factor, RAIDs are still fre­quent­ly used in server en­vi­ron­ments. RAIDs are designed to optimize data security and per­for­mance. In the case of RAID 10, in­di­vid­ual levels are merged with one another in order to combine their strengths.

But which levels make up RAID 10? What are its pros and cons and how does a RAID 10 system actually work?

A RAID 10 is a com­bi­na­tion of RAID levels 1 and 0, whereby multiple RAID 1 systems are combined with a RAID 0 setup. That’s why RAID 10 is also fre­quent­ly referred to as “RAID 1 + 0”. RAID 10 always consists of at least four hard disks.

Following the basic principle of RAID 1, in a RAID 10, the files are mirrored too. That means, data is du­pli­cat­ed to at least two disks in order to guarantee re­dun­dan­cy. At the same time, data is striped which means that data blocks are broken down into in­di­vid­ual stripes and stored on the disks in a RAID 10. Storage media within a sub­or­di­nate RAID 1 group always have the same data status, which in turn differs from the data status of all other sub­or­di­nate RAID 1 clusters. Each group manages separate parts of the in­for­ma­tion, but that’s of no con­se­quence to ap­pli­ca­tions accessing the system.

The failure of hard drive in a RAID 10 system is com­pen­sat­ed for via sub-arrays, i.e. via RAID 1 because a mirror with the same data is available at all times. However, that means the maximum storage capacity of in­di­vid­ual drives is half that of the in­di­vid­ual drives. Because data is evenly dis­trib­uted, users benefit from higher read speeds when accessing them.

RAID 10 stores all data in duplicate. As long as one disk of a mirror pair is still running, the in­for­ma­tion stored is therefore safe – even if a data carrier fails. Data is only lost if all storage media in a sub­or­di­nate RAID 1 fail due to a defect or another reason. This is one of the main ad­van­tages of a RAID 1 + 0 system compared to RAID 0 + 1 where damage cannot be assigned to any specific disk of a RAID 0 subunit. That all means re­con­struct­ing data in a RAID 10 system is more straight­for­ward and faster.

One of the great strengths of RAID 10 is the output rate of the system. By striping the data, the in­di­vid­ual sub-blocks are available in parallel. If an ap­pli­ca­tion accesses the system, it can read from two or more disks at the same time and always receives several data from a single strip. Compared to a single hard disk which doesn’t have that option, read speed in RAID 10 is always optimized.

Combining striping and mirroring is fairly expensive. But there are a few other dis­ad­van­tages. For one, the approach restricts the maximum storage capacity available, since for each gigabyte of user data one gigabyte must be set aside for a copy of the data. Secondly, RAID 10 risks data loss if both data carriers in a mirrored pair fail at the same time. In this case, the failure of two hard disks cannot be com­pen­sat­ed for by striping (unlike the failure of two disks from different sub-arrays).

Security and speed are among the greatest strengths of hard drive con­nec­tions according to RAID 10. Because costs of the setup are high, usage should be con­sid­ered carefully. However, if the budget allows it, a RAID 10 is always a sensible solution if you want to store data that:

• Needs to be accessed regularly;
• Needs to be read quickly;
• Needs to provide a high level of security.

Typical areas of ap­pli­ca­tion for RAID 10 are therefore database and ap­pli­ca­tion servers, which depend on a high level of re­li­a­bil­i­ty and a good output rate.

Al­ter­na­tive RAID levels 0 (striping) and 1 (mirroring) each offer only one of the prop­er­ties of RAID 10, i.e. either optimized per­for­mance or increased re­li­a­bil­i­ty. However, RAID 5 and RAID 6 also provide re­dun­dan­cy and increased per­for­mance but do not consist of a com­bi­na­tion of other RAID levels. Find out more about the different levels in our extensive RAID level com­par­i­son.