Virtualization on Storage Explained
Virtualization of storage means the same as virtualization of servers: Instead of having one machine piled sky-high with disks, you partition one machine into several storage devices or spread it across the network. Mainframes have always used virtualization for applications; a mainframe can run several different operating systems simultaneously. Virtualized servers work the same way; one machine is partitioned into different virtual machines each with its own separate memory space. Virtualization is even moving into networking, as companies seek to replace proprietary hardware firewalls and switch with software-based ones running on virtual servers. As virtualization grows in popularity, it is only natural that companies look to virtualize storage as well.
One type of storage virtualization is called (SAN). Storage area network (SAN) divides storage into arrays. Each array is divided into physical disks. The arrays are divided into logical units (LUNs). Virtualized SANs lets the storage appliance decide how to divide up the disks into arrays instead of having the administrator deal with all the physical device details themselves. The administrator requests LUNs and the appliance decides how to assign disks and spindles to arrays to meet the requested storage requirements. A “spindle” is a platter or the round piece than spins around inside a disk. The SAN can be expanded by adding another tray of disks and the SAN virtual controller decides how to use them.
Marketing people tend to take an old idea and give it a new name to draw interest and boost sales. Cloud technology is genuinely new, even if its definition varies. Virtualized storage is new only in name. EMC’s Integrated Cache Disk Array is 20-years old. It does exactly what we described above by creating an abstraction layer on top of the physical device layout.
The Microsoft Distributed File system (DFS) has been available for a dozen years. A virtualized abstraction layer sits between the user and the actual storage devices. The desktop is mapped to drives that are spread across the network. The metadata for the distributed file system is stored in Active Directory. Metadata includes information about where the actual device is located.
RAID too is a type of virtualization. In its simplest configuration, RAID 0 and RAID 1, two disks are presented as one. One mirrors the other.
The concept of “virtualization” is really nothing new with regards to storage. The impact on the organization is that storage vendors have built more intelligence into their controllers to provide them virtualization-like capabilities that lessen the need to configure SAN storage manually. This provides on-the-fly configuration changes based on changing storage requirements and changes in network performance. In their drive to jump onto the virtualization bandwagon, the vendors are continually improving their products, so it is important to keep abreast of changes.
