Storage Virtualization

Storage has long been a "first class" topic such as "the network." To this contributed significantly the introduction of technologies such as RAID (Redundant Array of Independent Disks) (Patterson, 1989) that promoted performance and/or tolerance to faults at (comparatively) low cost – it should be noted that in Patterson's original article I meant Inexpensive. With the introduction of RAID, disk lockers ( disk arrays), external to the "system" appeared, and since the SCSI protocol supported the interconnection of disks (Logical Units) to multiple adapters and, from there, to multiple systems, the concept of storage network appears naturally.

The logical disk appears as an elementary form of virtualization, since it "agglutinates" elements from different physical disks but is presented to the system as a single disk – to the point that it is impossible for the operating system itself to distinguish it from a physical disk.

In a storage network that connects multiple disk cabinets and servers (hosts), the former offer volumes (or logical disks) accessible by unique identifiers, and the latter take possession of those volumes (typically each host has exclusive use of one or more volumes) and format them, in them by installing file systems (e.g., ext3, XFS, NTFS), which they then use in the most convenient way. This storage network concept is known as SAN - Storage Area Network.

The most common form of SAN uses a Fibre Channel (FC) infrastructure, which means that both hosts and disk arrays have FC interfaces, and on the network there are FC switches to which hosts and arrays (and eventually other switches) connect. Another technology that can be used in SANs is Ethernet: in this case hosts, disk arrays, and switches have Ethernet interfaces, and the transport protocol used is Fibre Channel over Ethernet (FCoE).