By
Bryce Mackin
Network World, 09/09/02
ISCSI is a block-level storage
protocol that lets users create a separate storage network
using Ethernet. ISCSI uses Ethernet as a transport for data
from servers to storage devices or storage-area networks.
Because iSCSI uses Ethernet, it doesn't suffer from some of
the complexity and distance limitations that encumber other
storage protocols.
The iSCSI protocol puts standard
SCSI commands into TCP and sends those SCSI commands over
standard Ethernet. An iSCSI SAN consists of servers - with
an iSCSI host bus adapter (HBA) or network interface card (NIC)
- disk arrays and tape libraries. Unlike other SAN
technologies, iSCSI uses standard Ethernet switches, routers
and cables, and the same Ethernet protocol deployed for
communications traffic on LANs (TCP/IP). It can take
advantage of the same type of switching, routing and cabling
technology used for a LAN.
Because iSCSI uses SCSI commands,
relying on Ethernet only to transport the SCSI commands,
operating systems see iSCSI-connected devices as SCSI
devices and are largely unaware that the SCSI device resides
across the room or across town.
Most components inside these iSCSI
devices are very familiar to network professionals, including
RAID controllers and SCSI or Fibre Channel drives. The only
added feature is the iSCSI protocol, which can be run on
standard NICs in software or on specialized iSCSI silicon or
HBAs that off-load the TCP/IP and iSCSI protocol.
ISCSI is built using two of the most
widely adopted protocols for storage (SCSI) and networking
(TCP). Both technologies have undergone years of research,
development and integration. IP networks also provide the
utmost in manageability, interoperability and cost
effectiveness.
When SCSI is mapped to TCP/IP through
the iSCSI layer, SCSI is freed from its parallel bus
structure, letting users extend network storage using
Ethernet.
SCSI traffic at the application layer
moves to the physical network interface (either an NIC using
software or an HBA) then out over the Ethernet-based SAN to a
storage device, where the SCSI command is decoded and
presented to the storage operating system.
There are other components that can be
added to an iSCSI SAN to facilitate inclusion of other storage
protocols. These devices bridge the iSCSI protocol to Fibre
Channel and SCSI devices. Bridging products let users deploy
an iSCSI SAN without requiring retrofits to the entire storage
infrastructure. This means that storage can be transitioned to
become iSCSI-compatible, and the existing storage can stay in
place.
This bridging functionality lets users
preserve their existing investment while migrating to an
IP-based SAN. In addition to the bridging function, some of
the protocol-conversion devices also can provide switching
capability. This means that these devices can bridge iSCSI to
another protocol and simultaneously route the command to the
appropriate device.
An Internet Engineering Task Force
working group created the iSCSI standard. This group has moved
quickly to offer a universal standard. However, the
organization's job is not done yet. Ratification of the iSCSI
standard by the entire organization is still in process but
quickly moving to completion. Most are expecting ratification
of the standard this year.
Devices such as iSCSI HBAs, iSCSI disk
arrays, iSCSI tape and iSCSI to Fibre Channel/SCSI bridging
products are available today, and more will be available
throughout this year. Most of the products available today are
prestandard, but all have undergone rigorous interoperability
testing to ensure that they work in accordance with the
standard.
ISCSI provides IT managers with a new
option for solving increasing storage demands. By delivering
the advantages of SANs using a familiar and ubiquitous
Ethernet infrastructure, iSCSI unifies network and storage
deployment, making SANs available to the masses.
|
