Disk Configurations

Disk configurations - Logical volumes

When you add one or more hard disks to a computer system, or replace disks in a system, you have several options for the configuration of those disks. Each disk can appear to the computer as a single disk. A single disk may also be partitioned into several volumes with each volume behaving like an individual physical disk. Multiple disks may also be "spanned" so that they appear to be a single disk. Let's look at these different arrangements.

Disk volumes

Every hard disk has a volume structure that lists the size of the available space, and maps out how the data is broken up and stored on the disk. The volume structure determines what a computer "sees" when a disk is connected. In most cases, a single device will show up as a single volume, although that's not always the case. A single device could be partitioned into multiple volumes, and it's even possible to have multiple devices act as a single volume through drive spanning. (More on both of these below.)

Partitions

The most common volume structure for a single hard drive is a single partition, where to the computer the drive appears to be the one and only drive. It's possible to partition a drive so the computer sees it as two or more volumes. Partitioning separates sections of the drive from each other so each has its own volume information and directory.

Should I partition my drive?

While it used to be pretty common for advanced computer users to partition drives for performance reasons, that's less common now. We generally recommend that there's no reason to partition a drive, except for certain backup tasks.

Some backup arrangements like to have an entire volume set aside. Apple's Time Machine, for instance, will not let you use a single volume to backup more than one computer. If you want to send Time Machine backups from multiple computers to a single drive, you'll first want to make a partition on the drive for each computer. This is a reasonably common arrangement for bootable backup software, and it's the most common reason to partition drives.

Hard drive formatting

Hard drives may be formatted in several ways, depending on the hardware and operating system.

Macintosh

For Intel Mac computers, drives should be formatted as GUID Partition Table in order to be bootable. An Intel Mac can read a drive formatted as Apple Partition Map, but can't boot from a drive with this format. An Intel Mac can also read a FAT32 drive (a PC format that is also used on camera memory cards).
For Power PC Macs, (G5 and earlier), the drive must be partitioned as Apple Partition Map in order to boot from it. A PowerPC Mac with operating system later than 10.4 can read a GUID- formatted drive. PPC Macs can also read FAT32 drives.

Both Intel and PPC Macs can read from Windows NTFS-formatted drives, but can't write to these drives.

PC

PC drives should be formatted as NTFS, unless you know you need to share the drive with someone on a Mac. You can use FAT32 for more compatibility, but keep in mind it's a very old file system. For instance, FAT32 does not support files any larger than 4 GB. Windows computers can also be made to read Macintosh formatted drives (Apple Partition Map or GUID) by using the program MacDrive.
Read more about formatting drives for Mac and PC in this section

JBOD

The acronym JBOD stands for "Just a Bunch of Disks", and it refers to a configuration of multiple disks that retain their identity as individual disks. Each physical disk is a logical disk, as shown in Figure 3. It is the simplest arrangement for multi-disk storage, and the easiest to configure, upgrade, and repair. JBOD is a good configuration for photographers to use as primary storage for a couple reasons.

Figure 3 In a JBOD disk arrangement, each disk shows up as a single disk.

• It’s simple to purchase and set up. Making an informed choice on the best RAID implementation, on the other hand, is not something that should be done without a good understanding of the hardware and software under the hood. Few photographers are well equipped to make this decision properly.
• It’s generally much quicker and easier to upgrade capacity with JBOD. A traditional RAID setup must be completely rebuilt (all data moved off, all new drives installed) to upgrade storage size. This makes the upgrade process a difficult, time-consuming, and possible dangerous process. With JBOD, you simply add another drive or replace one drive with another drive of larger capacity.
• It’s the most economical configuration.
• Although RAID is theoretically safer, we find too many people who have lost their entire RAID systems for us to be convinced that it is practically safer, at least with consumer-level RAID.
• The amounts of energy used and heat generated are reduced in a JBOD set-up, since only drives that are being accessed need to “spin up.” Drives housing older material can remain at rest when unneeded.

RAID

RAID is an acronym that stands for Redundant Array of Independent Disks. In a RAID setup, several drives are configured together to act as a single volume. The computer sees multiple drives as a single device, as shown in Figure 4.

Figure 4 In a JBOD setup, each physical drive also apears as a logical drive.

This can be desirable for reasons of speed and/or redundancy. You can control your RAID setup either by using a hardware controller (such as a PCI expansion card) or by using software.

You can configure RAID systems where data is written to multiple drives with redundancy. This means that if a drive fails, all the data can still exist on the device. These mirrored or parity RAID setups can be a good choice to store working files in a mission-critical situation, such as a studio with lots of deadline pressure. And this kind of RAID is essential in a corporate environment where 24/7 uptime is necessary.

You can also configure RAID to produce significant gains in speed of throughput by reading from or writing to multiple drives simultaneously. This is useful in situations where there is a need for lots of throughput, such as in video editing. This can also help a photographer who routinely works with very large image files.

Let’s look at the various flavors of RAID and see what they do.

RAID Level 0 (striped RAID)

Treats several drives as a single drive in order to speed up reading from and writing to the drive. Basically, it breaks up a file into several parts and writes those parts simultaneously to multiple drives at once. While it increases speed, it also increases risk. A failure of one drive destroys all the data. RAID 0 is also used to create a single large volume. We generally do not recommend the use of RAID 0.

RAID Level 1 (mirrored RAID)

Writes the same data to multiple drives to help ensure seamless operation in the event of drive mechanism failure. Mirrored RAID is appropriate for operations that truly need 24/7 functionality, or for files that are works in progress and might represent a whole day’s work. Even with mirrored RAID, you still need to have a set of offline backups because mirroring does not protect against some of the most common causes of data loss (theft, virus, and power surge). It also provides no protection against human error, such as accidentally erasing or downsizing a file.

RAID Level 5

Provides a combination of striping and mirroring, resulting in a large single volume with redundant protection. A Level 5 system should be able to survive the loss of any single drive since a parity file (a copy of the data) is written to each of the other drives. Level 5 is probably the most popular current choice. The parity file is a compressed copy of the data; it can store the information in a small space, but requires computation to create.

RAID Level 6

An increasingly popular choice for enterprise-level systems since it builds upon the capability of Level 5, plus adding the capacity to survive the loss of any two drives in the system. This configuration is typically found in corporate environments.

RAID 0+1 (also known as RAID 10)

A configuration with multiple sets of striped drives for speed and redundancy. In a four-drive RAID 0+1, a pair of drives is striped for increased speed. Those drives are then mirrored to the other pair for redundancy. This is faster than RAID 5 or 6 since there is no parity file to be computed or decoded, but it is not quite as efficient.

Which RAID?

Okay, so despite all the above cautions, you want to set up a RAID. How do you go about it? Let’s take a look at some considerations:

Why do you want it?

Good reasons are that you need the speed, a single large volume, or that you need the 24/7 uptime. “It’s just easier to deal with” is probably not the right reason. Know if you are shopping mainly for speed, size, redundancy or some combination.

Hardware or software RAID?

You can use system software (or other third-party software) to span the drives and make a RAID from multiple drives, but we suggest you don’t. Instead, get a hardware-controlled RAID device that is designed specifically for that purpose. Remember, an error in writing the files can kill the entire RAID. In general, high-quality, dedicated devices do this better than software. More than one computer can also more easily access a hardware RAID.

Which hardware RAID?

There is a difference between hardware controllers. There are cheap ones from companies you never heard of, enterprise-level devices from well-known brands, and everything in between. Wiebetech is a company that makes, sells, and supports its own line of hardware RAID boxes (shown in Figure 5). Keep in mind that if the RAID card fails, it’s only remotely possible to access the data on the drives if you can get the identical RAID card as a replacement.

Figure 5 If you are going to get a RAID setup, consider a hardware RAID, rather than a software RAID.

If this sounds difficult to determine, you might want to think about working through a local reseller who can take responsibility for the purchase. Of all the purchases outlined in this website, this is the one that Joe Photographer is probably least able to do intelligently on his own.

Drobo

The previous section might have made you nervous about getting a RAID setup. There is a consumer-level option to consider if you’re looking for a fault-tolerant storage device: Drobo. Drobo’s goal is to offer many of the advantages of RAID with some significant improvements, all in a device that is a simple storage appliance.

Figure 6 Drobo has made implementation of drive spanning easy for anyone to implement.

At first glance, Drobo acts like a RAID—it spans up to four drives to make one logical drive. It offers the same kind of parity protection you get with RAID 5, where the data can survive the loss of any single drive (or, if the drives are large enough, the loss of any two drives, like in RAID 6).

However, Drobo offers some things RAID does not. It can intelligently make use of drives of different sizes (unlike RAID, where all drives are chopped down to the size of the smallest drive). When you need to upgrade the capacity of your Drobo system, just take out the smallest drive and replace it with a larger one. The data will be swapped around, making the best use of the new drive space.

Drobo also has some very simple “gas gauge” lights that show you how full it is. As the disks get full, it will show you on the front of the box itself, as well as in the Drobo Dashboard software that comes with the unit. It connects to the computer by USB2 or FireWire 800 connection, and offers an add-on Ethernet connection for use as a NAS drive.

Drobo also has some intelligent power consumption features. It goes into a very low-power-draw sleep when not connected or not used, then wakes up when it’s needed. And because it is a standardized item, the controller card is likely to be available in the future (in the event that the one in your unit dies and you need to retrieve your data).

There is one important disadvantage, however, compared to something like RAID 5. Drobo is not very fast for reads and writes. The newer FireWire unit adds additional speed over the original Drobo, but you should probably not consider it as primary storage for use in an environment where high throughput is important. We recommend using one as a working files backup that is accessed in the middle of the night.

DroboPro

In April 2009, Data Robotics introduced DroboPro, a new version of the product that adds capabilities normally found in enterprise-class storage hardware. DroboPro adds support for up to 8 drives, and includes a faster processor for better throughput. DroboPro also makes use of the iSCSI connection protocol.

Figure 7 The DroboPro can span up to 8 drives for up to 16 TB of storage.

NAS

NAS comes in two flavors: self-contained boxes about the size of a FireWire hard drive, and NAS servers that are effectively small, fully functional servers with connection ports for monitors, keyboards, and external drives. The NAS servers provide an easy way to add a server for a low cost if you don’t already have a computer to turn into a server, but it has limited capability to recover if problems do arise.

Although we generally don’t recommend using NAS as primary storage for professional photographers at the present time, it can be a good solution for someone who wants to keep backup storage online far away from the primary storage. Since you can connect it by Ethernet cable, you have the flexibility to move it farther away than any other kind of wired storage.

Some NAS products, such as the HP MediaSmart server run on Windows Home Server, and can handle a number of different kinds of files. These devices can serve audio and video files as well as simply storing image files. They can also be used to make files available over the Internet without having to do a lot of custom configuration.

Figure 8 Stand-alone devices like the HP MediaSmart Server can provide additional storage or backup over a network without having to set up a dedicated server computer.