After we ran an article describing IBM’s efforts in I-SCSI, many
of our readers asked for more information on exactly what I-SCSI
was and what it would be used for. This short article at Network
Storage Forum should answer many of those questions.
“I-SCSI uses warm and fuzzy TCP/IP networks to carry
data. With I-SCSI, data transfers and SCSI commands get packaged
(or encapsulated) inside IP packets. Instead of having to go
through a lot of hassle and expense to put together a fibre channel
network just for storage, with I-SCSI you can use the tried and
true old network standby, Ethernet. Besides, you probably already
have an Ethernet network installed in your data center. So, forget
the fibre channel. Using SCSI commands sealed away in the IP
packets lets you use existing storage management software with
little modification. Great. You won’t have to spend hours learning
some new storage management mumbo jumbo.”“By using I-SCSI, you’ll get to save a lot of dollars on long
distance links between pools of storage. Some of those exotic high
availability strategies require data centers to be mirrored to
locations several miles away. The cost of running a fibre channel
connection over such a distance will make you think twice about the
economics of doing this. On the other hand, I-SCSI enables you to
use existing network connections without any modifications. You can
use I-SCSI over short distances, too. Keep in mind, Ethernet
technology is moving to gigabit Ethernet, which can support up to
125 Mbytes per second.”