Using iSCSI for Network Storage Solutions

I'm confused by the term Internet SCSI (iSCSI). What is iSCSI and what's its most important use?

The iSCSI protocol encapsulates standard SCSI commands into Ethernet packets for transport over TCP/IP networks. Thus, iSCSI lets you apply network storage solutions in an open environment without using any additional framework, such as Fibre Channel. You can construct a SAN by connecting servers via Ethernet cards (usually Gigabit Ethernet) and iSCSI software initiators or iSCSI host bus adapters (HBAs) and drivers. The servers see the storage as local drives. Like all SANs, you can expand the storage capacity without taking the existing storage offline, so iSCSI is an ideal alternative to more expensive storage solutions that have additional substrate requirements, such as Fibre Channel cards and cable.

The iSCSI protocol has been in development for several years. Because of the exacting nature of the SCSI standard, the adoption of iSCSI has been controversial. For example, Sun Microsystems has argued that the lag time or latency of TCP/IP limits the data transport rate and thus restricts the development of the iSCSI protocol, and indeed, the development has been fraught with many difficulties. However, developers have addressed and to a large extent solved these problems. Companies such as HP and IBM, as well as standard SCSI HBA suppliers, have been the primary proponents of iSCSI.

A major component of iSCSI is its ability to transport, in order, SCSI commands over a network. This functionality hasn't been easy to develop, but the outcome looks promising, with many companies now providing reasonable prices for iSCSI SANs that offer a local storage solution easily shared by network servers. iSCSI is an attractive alternative for companies that don't want to contend with new hardware alternatives to standard, well-established protocols.

—Bob Chronister