By clicking ‘Accept’, you agree to the storing of cookies on your device for an enhanced experience as well as analytical and commercial purposes. To learn more about how we use cookies, please see our privacy policy.

RAID storage: A guide

The concept of aggregating your data across several hard drives is nothing particularly new. Early arrays of hard drives first appeared on the market during the late 80’s and early 90’s and were used in large commercial data storage environments. In the last few years RAID drives have become better understood and have been adopted in more mainstream environments. For those who aren't so familiar, let us help you catch up.

The acronym ‘RAID’ is short for ‘Redundant Array of Independent Drives’ and is a way of combining multiple, physical drive units into one, logical unit. The main purpose of RAID is increased reliability, performance, data redundancy and fault tolerance. A data-loss scenario for most companies, even small businesses, would be a disaster. RAID can ensure your data is intact in the event of hardware failure – a very attractive attribute for businesses and commercial enterprises when considering how to store large volumes of business-critical data.

In some RAID setups, its possible to have multiple drive failures without losing any data. Compare this to a regular desktop computer system which only has a single drive installed. Should that drive fail, then all of the data held within it will often be lost, too (we’d always recommend a backup is held on another form of media, but even now, that is often not the case).

There are three principle types of RAID. These are known as “mirroring”, “striping” or “striping with parity”.  Each offers different benefits for different usage scenarios.

Mirroring is the simplest form of RAID, which simply stores the same data across two physical hard drives. This provides the simplest level of redundancy.  If one drive happens to fail, then you can still access all of your data as the other is a complete ‘mirror’ of it. The downside is that the total capacity of your storage unit is halved, i.e, If you have two 2TB units (4TB in total), you would only be able to store up to 2TB of files.

Striping distributes your data across several hard drives. This takes considerably more hardware processing power however, so the read & write performance can be slower than you’d get from a mirrored array. On the up side, you get the full, aggregated volume of data which is equal to the hard drives in the RAID setup. For example, if you had an 8TB RAID, made up of four 2TB drives, you’d have the full 8TB available for storage. This form of RAID storage doesn’t in itself, offer any tolerance to a drive failing, but can offer large amounts of data storage in what appears to the end-user, as a large, single volume.

Finally, there’s “striping-with-parity”. This is similar to how striping works, with data distributed across several drives, but a ‘parity bit’ (sometimes called a check), is stored on another drive in the RAID setup. The benefit of this form, is that data can be reconstructed if one, or even sometimes more, drives fail (depending on the number of drives in the configuration). Whilst this allows for some fault tolerance, the downside is loss of performance, as there is often a large amount of data processing being carried out by specialist hardware called a controller card. Also, the total amount of data you can store, is less than the aggregate of the drives together, i.e, on a 10TB setup, made up of five, 2TB drives, you’d end up with in the region of 7TB of file storage. This is due to at least one of the drives being used entirely as the parity-storage drive. Also, all of the other drives would also contain smaller pieces of the parity data, too.

You may see a lot of talk within RAID setups about ‘levels’, as the types of storage above are defined by level numbers. Here are some simple explanations of these levels in practice -

RAID Level 0

This is the simplest form of data ‘striping’, with data split between two drives (though it could be 3,4,5 or any number of drives).  A drive appears to the user as one, contiguous unit but, offers no fault tolerance or security. It does, however, offer faster performance but at the cost of reliability and vulnerability to data loss.

Many budget, but very large capacity hard drives may well be using RAID 0 by striping many cheaper, smaller capacity drives together to give the appearance of a large, single drive unit. Some would argue that RAID 0 isn’t really a valid form of RAID at all, as it doesn’t offer any fault tolerance, which is a fundamental benefit of RAID setups to many experts.

RAID Level 1

This is the purest, most basic form of data ‘mirroring’ as described above. It offers fault tolerance, but cuts the total available data capacity by half. Usually found in more 'consumer' level devices where two drives are linked together

This setup is useful in low data capacity environments but where some fault tolerance is necessary, possibly in graphic arts where there is a high turnover of data.

RAID Level 5

Level 5 offers striping-with-parity across several physical drives. Whilst this gives you excellent fault tolerance and security, it comes at the cost of performance and a reduction of storage capacity, due to the overhead.

This form of RAID is ideally suited to a business-critical data storage like back-ups, where speed isn’t a huge requirement.

RAID Level 10

A combination approach, linking RAID 0’s striping with RAID 1’s mirroring, usually with four physical drives, where more performance is required along with an amount of fault tolerance, at the cost of reduced capacity.

This particular setup, similar to RAID 5, would be best in a business-critical data storage situation but as the processing ‘overheads’ are considerably less, users with a preference for performance may find this a better option.

Planning ahead...

It is advisable to spend some time at the outset to think about your priorities – whether you place a higher priority on better performance, a larger amount of storage or drive redundancy. Considering which RAID setup would not only work well for you now, but at a later date, as it can be a nightmarish task to migrate your data to a different RAID level.

A common misconception with RAID is to think that a RAID-array doesn’t require a backup. In reality, the complete opposite is the case. Just because the drive and the RAID level you’ve chosen may have some redundancy doesn’t immunise it from other problems. Physical damage to the array (a leak, a flood, a power interruption etc.) could result in data loss, as could a theft of the drives for example. Many RAID products offer a means of plugging in a USB or Thunderbolt hard drive to back-up your data to for additional peace of mind.

There are many different ways to access data on a RAID unit. They can be directly attached to your Mac via a Thunderbolt interface or, if you are planning on sharing data across a network, some units have an ethernet. More advanced, corporate products may use a specialist fibre-channel connection, for improved data transmission speeds (note that this requires specialist hardware and network).

When you’re at the purchasing stage, factor in that your data needs will expand over time. You may consider a 10TB drive more than adequate now, but if you were to use RAID 5 as a format, then you are losing 2-3TB before you start, so choose capacity wisely and think into the future.

Finally, if you do not currently have a backup plan, consider how you’ll save your data to ensure you’re not putting all of your eggs in one basket. As previously stated, do not rely on the fault-tolerance of the drive alone. Purchasing an independent, large-capacity USB drive and a plan to ensure regular backups take place (and the back up unit being stored separately of the RAID storage, for obvious reasons..), just in case something happened to your RAID unit.

In conclusion, there is no ‘right’ or ‘wrong’ solutions here, with every business or user scenario requiring a different solution to address their data storage needs. With careful consideration and forward-planning, a suitable RAID solution is certainly a more progressive and considered way of storing large volumes of data securely for many businesses, schools and institutions.

Follow us on Twitter.
Email us at info@krcs.co.uk
Call us on 0115 9851797

Comment Below

Comments

    No Comments yet. be the first to comment.