Why RAID, a Technical Description

Why RAID?

Technology Brief

Why RAID?

SANpath: Increasing Performance, Eliminating Failure Points, Shared Storage for Multiple Servers.

Copyright  Dot Hill, 11/99 Rev. xx1

Why RAID?

Introduction

In short, a RAID system today refers to a data storage device that incorporates a group (or an array) of disk drives that are joined together for a particular purpose. This joining can be done electrically (by hardware) or logically (by software) or sometimes even both. In each case, the reason or goal for joining the disks involves providing a feature or function that cannot otherwise be achieved by a single disk drive alone.

RAID(originally defined as a Redundant Array of Inexpensive Disks)

However, the word “Inexpensive” became politically incorrect over the years, particularly for companies that charge big prices for their RAID systems! So today, more often than not, RAID gets conveniently redefined as a

(Redundant Array of Independent Disks)

In either case, the technical description is the same and is presented below in simple terms.

The three (3) most important benefits of RAID storage systems are:

RAID data storage systems typically provide performance that is many times faster than a single disk drive. The set of drives in a RAID system work together (often in parallel) in order to multiply their individual performance.

Performance comparison using approximate values:

Performance Range1
Single Hard Disk / 10-30 MB/sec.
RAID system / 35-350 MB/sec.

1Can vary significantly between brands and models.

RAID systems typically offer storage capacities that are much greater than that of a single disk drive. This generally stands to reason since multiple disk drives are usually incorporated into each RAID system. With few exceptions, each drive in a RAID unit adds more storage capacity to the system.

If a RAID system is not fully loaded with drives when purchased, it is possible to add additional drives later to increase its total capacity.

Note: When adding drives later to a RAID system, it is usually necessary to erase the whole unit and “reinitialize” or “reformat” it. (There are some exceptions to this on higher end systems using specialized software). In general, one should plan on backing up all of the data from the RAID system to tape, or another RAID unit, before expanding its capacity. After expansion drives have been added, and the RAID unit has been erased and reformatted, the backed up data can be restored to the RAID system. Operations can then resume, with more space!

RAID data storage systems offer excellent protection from failed disk drives1 (and other failed components too2). For many folks, this is the most important feature that a RAID system provides.

If a hard disk fails on a typical server or workstation computer, all of the data stored on it is usually lost forever. This can really disrupt a business, not to mention ruining someone’s day. Often times, this lost data cannot be replaced. When the data that’s lost is valuable, the consequences can be dire, and the losses (beyond the data itself) can be significant (i.e. lost time and revenue, lost reputation – particularly if the data belonged to a customer or business client, lost passwords and authorizations, etc.. Even certain kinds of legal liabilities can result). In fact, all manner of business and personal disruptions can result from failed storage devices (especially hard disk drives)! And, these components can fail at any time – without warning! The value of current backups is never more keenly felt than after this type of bad luck3.

In general terms, a RAID system stores data in a distributed way across its set of drives. Each drive saves a segment of the total data4. If one of the drives in the set fails, special codes (and compressed data) 5 on the surviving drives allow for the data on the failed disk to be “reconstructed”. The result is that no data is lost, even if a drive in the RAID system fails2.

What’s in a RAID level?

Speed / Storage Capacity / Protection
RAID 0 /  /  / NO
RAID 1 /  / less efficient / 
RAID 3 /  /  / 
RAID 5 /  /  / 

Table 1.1

Connecting RAID systems

Connecting a RAID system to a single computer is usually quite straight forward. In most cases, hardware RAID systems connect (and are prepared for use) in the same manner as standard external hard disks. Once it’s physically connected, the RAID unit can typically be partitioned and formatted with the same software that one would use for a standard hard disk. To most utility software, there should be no appreciable difference between a hardware RAID system and a standard hard disk (other than the larger storage capacity). Where RAID systems due frequently differ is in the area of remote management and status monitoring software. However, for single computer “point-to-point” installations, these areas of difference are generally optional and not mandatory for deploying a RAID system.

Virtually all RAID system suppliers provide documentation that describes the step-by-step instructions for connecting their hardware and the use of any special or recommended software.

With most computers, the RAID unit can be set to be the “boot” device. This allows the operating system, applications, custom user settings, drivers etc. to all benefit from a RAID system’s attributes. Computers using RAID storage can boot faster and survive disk component failures without losing data. For servers or workstation computers that must be “highly available”, RAID systems are a must.

Sharing a RAID system and its benefits

Figure 2 below shows an example of one way to share the benefits of a RAID system with multiple users on a network. Briefly, by connecting the RAID unit to the common server, each client can benefit from its extra capacity, speed, and very importantly – its data protection. This type of configuration also simplifies backup tasks by centralizing more client storage.

Notes:

1Doesn’t apply to RAID 0 systems. RAID 0 systems “do not” offer this data protection. RAID 0 is sometimes referred to as “nonredundant” RAID. See table 1.1 “What’s in a RAID level?”.

2Many RAID systems offer redundant (or backup) cooling fans, power supplies, sensors, controller boards, displays, and even control pads as part of their design. If one of these components fails, the other will typically take over automatically. Often, failed parts can be replaced without turning off the system. These are often referred to as “hot swappable” components.

3Regular, current backups are always recommended – even if one employs RAID systems. RAID systems are not a replacement for good backup practice and policy. Instead, they are added insurance against data loss and business disruption.

4This applies to RAID levels 2, 4 and 5. Along with the data itself, special codes5 and a compressed version of the data are also saved to each drive2. For RAID level 1, a full “uncompressed” copy of the data is stored on the other drive(s) and/or partitions within the RAID system. RAID level 1, while providing the same data protection as the other levels1, offers less “usable” capacity for an equal number of drives.

5Sometimes called parity or hamming code. Note that RAID level 3 stores these special codes and compressed data on a dedicated disk in the set instead of spreading it out among all of the drives.