iSCSI’s long-standing battle with Fibre Channel for control of the heavyweight SAN belt is far from over, but iSCSI may meet another foe within the next year or two. The proposed Fibre Channel over Ethernet (FCoE) standard would, like iSCSI, allow storage traffic to flow over ubiquitous, inexpensive, easy-to-manage, etc., Ethernet (see “Making the case for FCoE,” p. 10).
As noted in the article, when FCoE was first proposed back in April it engendered a good deal of controversy. iSCSI proponents argued that FCoE was little more than a defensive attempt on the part of Fibre Channel proponents to thwart adoption of iSCSI and to put up a road block on the highway to nirvana: Ethernet as the sole conduit for all IT traffic: storage, networking, and clustering.
Although it can be argued from a political or technology angle, the iSCSI vs. FCoE debate is not as controversial as it may seem. There will definitely be a place for FCoE-in large enterprises that have invested heavily in Fibre Channel SANs and for whom replacing Fibre Channel with iSCSI is, and will probably forever be, a ludicrous proposition. For those companies, FCoE is the best of both worlds. It enables the preservation of existing Fibre Channel investments (hardware, software, and expertise) while taking advantage of the skill sets and low cost associated with Ethernet (although it will be quite a while until “low cost” is associated with 10Gbps Ethernet, which is the presumed transport for FCoE).
However, FCoE will not make sense for small to medium-sized businesses (SMBs), or at least for the “S” part of “SMB.” For them, iSCSI is a very viable alternative, and the extra costs associated with implementing FCoE (specialized host bus adapters, switches, bridges, etc.) will be prohibitive. Besides, most of these companies don’t have Fibre Channel SANs, rendering FCoE totally irrelevant.
It’s pure conjecture at this point, but I see iSCSI continuing to make steady inroads at the low-end of the SAN market (at least among those companies that actually need a SAN), while Fibre Channel-augmented by FCoE and Ethernet-maintains its stranglehold on the high-end of the SAN market. The battle will continue to rage in the “M” portion of the SMB space, but even there I don’t see FCoE making any headway because of the increased costs.
The more things change, the more they remain the same. But if you’re still craving controversy, in the time frame that FCoE will see the light of day (2009 or later) I predict that InfiniBand could upset the status quo and fuel the flames of controversy.
In the end, however, the battle for SAN supremacy may have little to do with protocols: It will be a battle for which IT group controls which-or all-types of traffic: the storage, networking, or server groups.