• United States

iSCSI vs. Fibre Channel

Apr 13, 20043 mins
Data CenterSAN

* The march to networked storage

While it’s a fact that most storage is still attached directly to the server that accesses it, is also true that storage continues to move at a steady pace onto the network.  In all but the smallest IT environments most newly purchased storage – be it JBOD or array – is going to be either network-attached storage or storage-area network.  Networked storage is clearly where the investment is being made.

When storage moves to a NAS device, it is usually because of a desire for either economy or simplicity.  Many cheap NAS devices are available now, although it is not at all clear that putting lots of small NAS boxes on a LAN is truly economical when it comes time to manage all the devices. Larger NAS boxes may not be at all cheap, but they certainly are versatile and manageable.

When storage moves onto a SAN, it does so for two main reasons: performance is better, and it is easier to manage.  Management of course means many different things, but eventually the definition always has something to do with optimizing the use of assets, and by doing so, optimizing IT investment. In other words, if a SAN can make all data available to all users (through the switches, it does), and if a SAN can make sure that all disks get used to their maximum (a storage pool virtualized across all the storage hardware does this), buyers have found a compelling argument in favor of concentrating their purchasing on a SAN.

When it comes to installing new SANs, IT managers often ask whether these should be based on Fibre Channel or iSCSI.  Both technologies obviously have their proponents, and both have their detractors. 

A sensible answer to this question has to do with a lot more than the technologies involved.  It also should consider what sort of expertise the site currently has.  If the IT team is neck-deep in Fibre Channel experience, then it probably makes very good sense for them to continue using fiber.  On the other hand, if they are new to SANs but have lots of Ethernet experience in-house (just about everyone qualifiers here), iSCSI would seem to be a relatively painless choice.

At many sites however, the determination of the iSCSI vs. Fibre Channel question is ultimately going to be resolved along the lines of who holds the purse strings.  If purchasing authority lies with the corporate IT department – who often have deep understanding and significant investment in Fibre Channel, Fibre Channel remains the first choice.  If purchasing authority has been deployed out to a business unit, to a remote group, or to a midsized IT shop that does have much experience with fiber and doesn’t have the wherewithal to invest in a full-scale Fibre Channel build out, iSCSI is a frequent choice.

If all this is right, and I am pretty sure it is, then there is an interesting point here for the SAN vendors.  It is this:  the iSCSI vs. Fibre Channel decision has a lot less to do with the size of the company than it does with who is doing the buying.  That being the case, iSCSI SANs may have a much wider play than the midsize businesses at which they are most frequently targeted.