Mirrored and concatenated RAID options in Mac OS X

Apple's Mac Pro computers have the capacity of holding up to four internal hard drives, and all Mac models support the use of external drives through either USB or FireWire. This allows for multiple drive configurations that include the option for various RAID setups to increase productivity and storage efficiency based on the type of RAID used.

Some people have wondered about the RAID options available on the Mac and what might be best for their purposes, such as Apple discussion poster "Chulo01" who writes:

"Wondering if anyone used [LaCie 3TB Big Disk or 2Big drives] and what the advatages and disadvantages of it are as supposed to building a RAID 0 from scratch. I'll be using the eSATA connection and I haven't found anything on speed differences.

The portability on the Lacie is what I'm leaning towards and the multiple interface, but I'm curious to know if I'll be sacrificing speed for this since I'll using it to edit video on Final Cut Pro."

RAID in OS X

A RAID array, or Redundant Array of Inexpensive (or Independent) Disks, is the use of multiple storage devices together to create one device that has benefits over each independent unit alone. The benefits depend on the RAID type being used, and of the many RAID options Apple supports four in software.

• Striped RAID set (RAID 0)
  This is the most basic RAID level for concatenating two or more disks and achieving both larger storage size and faster read and write speeds. It evenly distributes data across all member disks, allowing all the member drives to be used at once during a read and write routine. The benefits of this setup are it is the fastest RAID option and makes all the space on all drives available for use; however, as the number of drives grow the risk for data loss grows with the increased probability of one drive failing. Unlike more advanced versions of RAID 0 (Raid 4 and RAID 5), RAID 0 does not store data parity information so if one drive goes out you cannot rebuild the array and the data is lost. It can be more risky, but it's fast and as such is great for use as a "scratch" disk to hold temporary files and other work flow in high disk access tasks such as rendering movies.


• Mirrored RAID set (RAID 1)
  Unlike RAID 0 that makes the most of speed and storage space, RAID 1 is a mirroring option that uses redundancy to ensure both data and work flow security from hardware failure. If one drive in the array goes out, the other drives will continue to provide same copies of the data to the user without interruption. In most situations, RAID 1 is only used between two disks because it can easily be rebuilt in the event of a drive failure, and having two copies is usually more than enough. Any more would be an overly redundant use of mirroring, and as such is usually considered a waste of drives; however, it is additional security for those who wish for it. Mirrored RAID setups are useful in situations where continual work flow is critical, such as the boot drive of a server. Keep in mind that mirrored RAID arrays will only be as large as the smallest disk used in the array, so if a 250GB drive and a 400GB drive are mirrored, the resulting array will only be 250GB and the remaining 150GB on the 400GB disk will be lost.


• Concatenated Disk Set (JBOD)
  Both striped and mirrored RAID setups require the member drives to be the same size so no space goes unused; however, if you have multiple disks of varying sizes you can link them together to create one big disk with a concatenated "JBOD" array (JBOD stands for "Just a Bunch Of Disks"). It will not benefit from speed increases or data redundancy, but it will allow for more efficient use of the space offered by all drives working together. Data is not distributed evenly across each member drive, but instead one drive is filled, and then the next one and so on as data is added to the array. If one drive is slower than another, data access will be slow only when that drive is being written to or read from. The most common use of this drive type is when people have multiple older drives of different sizes that they can put in one computer.


• Logical Volume Group (the hidden option)
  Logical volume groups are a way to manage existing partitions in a RAID-like behavior instead of handling RAID on the device level. This lets you select any number of partitions from various devices and group them together to behave like a concatenated RAID set. The benefit of using a LVG is you can keep the data on a currently partitioned disk intact while setting up a RAID array between unused partitions on the same disk and others. This option is available once you've started creating the RAID array (see below).

The setup

RAID can be used on any locally mountable storage system; as long as two or more hardware devices are physically present, they can be used in a RAID setup. While the use of internal hard drives is recommended over the use of external USB and FireWire devices, these less robust devices are an option; however, keep in mind that RAID arrays will keep your hard drives active, and many USB and FireWire drives are put in cheap enclosures that do not dissipate heat well. As such, the use of these devices may increase the probability of failure if put in high-use situations such as RAID arrays.

To set up the array, first attach all devices and then launch Disk Utility and select the "RAID" tab in the window to the right of the device list (this can be done from the OS X installation DVD if you wish to install OS X to the array afterward). Name your RAID array and choose a format and RAID type.

For optimal performance, you can set the block size for the array. In most cases a 32K or 64K block size should work fine, however, for smaller files such as office documents a smaller block size will make more efficient use of disk space, and for movie, audio, and other large media files a larger block size will be faster. To set the block size, click the "Options..." button and select between 16K and 256K for the block size (for more information on block sizes, see this article).

If you are using a mirrored RAID set (RAID 1), you have the option to automatically rebuild the array in the event of a drive failure. We recently discussed how to manually do this in our tutorial on rebuilding software RAID mirrors, but disk utility will attempt to do this automatically if you have both a spare drive installed (see below) and this box checked. Rebuilding will only work on mirrored RAID arrays and those that store parity information (RAID 4 and 5, which are not supported in software).

Under the RAID sets window, click the " " button and an array set will be added with the information provided so far. To change the RAID type to an LVM, right-click the array and choose "Logical Volume Group" from the menu.

Drag all member disks or volumes (for LVMs), including any you wish to be "spare" drives, to the newly created array, and designate any spares as such by right-clicking it and choosing "spare," ensuring in a similar fashion that the others are each listed as "member." Then click "Create" and the array will initialize. All data on the member drives will be erased and the RAID array will take some time to set up. When it is done, the array will be present to the system in a new volume that will mount and be ready for use.

Other options: Hardware

While we have covered how to set up RAID arrays in software, and while this is by far the cheapest way to set up a RAID array, it is far more limited than hardware RAID solutions. Some things to consider about hardware solutions that may make them more appealing are:

1. More options
   Hardware solutions have the option to set up RAID 0 through RAID 5 array types, including combinations of each of these. This allows for the use of parity information in concatenated and striped RAID arrays that allows for recovery of data and continued workflow when a drive fails in the array.


2. Speed
   Having dedicated controllers manage the arrays makes them much faster on hardware RAID setups than those managed through software. When connected through a high-speed fiber channel they can be much faster than most independent hard drive solutions available today.


3. Hot-swappable drives
   Hardware RAID systems usually support hot-swapping of drives, which allows for easy recovery in the event of a drive failure. You can quickly remove a drive, put in a new one, select it for use with the recovery, and have the system rebuild the array.

Hardware solutions are by far the more preferred RAID options, but can cost thousands of dollars. We recommend against using cheap RAID solutions for important files since they can fail and may not support more substantial RAID solutions such as RAID 4 and 5, but the market is evolving and more robust RAID solutions are making their way into small form factors. Still, be sure to research a drive's reliability before putting all your data on it, and have a solid backup solution when you are using concatenated RAID setups.

Resources