A disk array is a hardware element that contains a large group of hard disk drives (HDDs). It may contain several disk drive trays and has an architecture which improves speed and increases data protection. The system is run via a storage controller, which coordinates activity within the unit. Disk arrays form the backbone of modern storage networking environments. A storage area network (SAN) contains one or more disk arrays that function as the repository for the data which is moved in and out of the SAN.
Related Terms: LUNs and RAID
Logical Units: The data within a disk array is organized into Logical Units (LUs). Small Computer Systems Interface (SCSI) I/O commands are sent to a storage target and executed by an LU within that target.
LUN: A Logical Unit Number (LUN) is a unique identifier that is used to distinguish between devices that share the same SCSI bus. Commands that are sent to the SCSI controller identify devices based on their LUNs.
Redundant Array of Inexpensive (or Independent) Disks (RAID): RAID employs two or more drives to improve performance and fault tolerance. RAID enables the storage of data in multiple paces to protect the data against corruption and to serve it to users faster. There are various levels of RAID. RAID 0, for example, provides data striping, whereby data is spread out across many disk drives without having any redundancy. This is a good way of improving performance. RAID 1, on the other hand, has disk mirroring so that all data is written to two physical disks and the complete data exists in two places; if one disk fails, the other one is available. This is a good way to speed reads, but the downside is that it slows the write process. RAID 10 is a combination of RAID 0 and 1. There are also further types of RAID, like RAID 5 and RAID 6.
Disk Array Vendors
EMC established itself in the 1990s as a disk array specialist. However, there are many other vendors offering disk arrays, including NetApp, IBM, Dell, Oracle, HP, Hitachi Data Systems, Compellent, Pillar Data Systems, Xiotech, Fujitsu, NEC, Data Direct Networks, Huawei Symantec, Nexsan, SGI, Overland and Infotrend, among others.
At the lower end of the market, there are a wide range of vendors offering relatively cheap disk arrays, including Buffalo, LaCie, Dulce Systems, G-Technology, Newer Technology and EMC's Iomega division. Vendors such as Dell, HP and Overland also offer inexpensive lower-end disk arrays.
Disk Array Technology
Disk arrays contain the many spinning disks where data is stored. In other words, a whole series of disks are arrayed inside. In such a grouping, many physical disks can be combined to form much larger logical disks. The storage controller can dictate that three physical disks of 100 GB be combined into a logical volume of 300 GB, for example. When you have hundreds of spinning disks in the unit, you can then create a few really huge logical volumes if desired.
The term disk array also applies to a wide range of different devices. At the high end, monolithic storage arrays are very large and expensive and come complete with enterprise-class features such as full redundancy. If one disk or other component fails, the system continues without pause. Thus enterprise-class disk arrays are designed to be highly available. They hardly ever go down. Further, they come with a large quantity of memory. Frequently accessed data can be stored in memory and served rapidly.
Enterprise disk arrays also make heavy usage of RAID. These arrays often contain two high-performance redundant RAID controllers to serve their large quantity of disks. Some RAID configurations mean that the system remains running if one disk fails. However, RAID 6 is an example of a type of RAID which can cope with two simultaneous disk failures without downtime.
In addition, there are a wide range of disk arrays to serve the midrange, small business and even the consumer markets. These days it is even possible to purchase multi-TB disk arrays for a few hundred dollars. However, these products lack sophistication, don't come with the redundancy and other features of enterprise arrays, and also are performance constrained. Whereas an enterprise system might be able to serve data efficiently to mainframe computers processing a dizzying amount of transactions per second, low-end disk arrays are mainly aimed at environments where only a few PCs or servers are accessing the data.
Over time, even the low-end boxes are incorporating RAID and basic data protection technologies into their arsenal. And as processor and connection speeds increase, they become more and more capable of achieving a reasonable level of performance.
- Storage Networking 101: Configuring Disk Arrays
- EMC VNX Disk Array Buyer's Guide
- IBM Storage Buying Guide
- RAID Storage Levels Explained
- The Future of RAID Storage Technology