Published: 12 Oct 2008
| End in sight for Fibre Channel?
The first chinks in FC's armor come as a result of iSCSI storage gaining wider acceptance. Back in iSCSI's early days, vendors offering those systems were careful not to step on the toes of the big storage guys and positioned their systems as complementary to--rather than competitive with--FC. That was a pretty accurate assessment for the most part, as initial iSCSI storage arrays had some shaking down to do and certainly couldn't match the credentials of FC storage. But I think that strategy by iSCSI vendors was partly a ploy to stay out of the sights of the big vendors while leading them into the iSCSI market. Premeditated or not, the plan worked, and you'd be hard-pressed to find a storage vendor today that doesn't have iSCSI in its product lineup.
The rest is history or at least history in the making. As we've seen in our Purchasing Intentions surveys (see "Economy and capacity at odds"), iSCSI isn't just a flash in the pan and it's more than just a low-cost box you toss in to fill out tier 3 or tier 4 storage environments. Nearly a third of the big businesses represented in the survey have iSCSI systems on the floor, and they're not being used just for low-priority apps. A good portion of these companies are running email and other mission-critical apps on their iSCSI gear, applications that were previously the sole domain of FC storage.
SAS drives represent another threat to FC. One of the knocks against iSCSI was its poor performance vs. FC storage, but SAS drives help to narrow that gap. SAS brings a bevy of benefits to the drive table, including good performance (15,000 rpm and 3Gb/sec), smaller form factors, scalability, co-existence with SATA drives and multichannel aggregation for even better performance. And the SCSI Trade Association's SAS development roadmap calls for 6Gb/sec drives in the next year or so, which will put SAS a step ahead of current 4Gb/sec FC systems but still trailing the 8Gb/sec FC products that are on their way.
The third threat to FC's hegemony bears its name: Fibre Channel over Ethernet (FCoE). FCoE attacks FC from the network side although, as its name implies, it's positioned as a complementary network architecture to FC.
FCoE is billed as a "uniter not a divider" because it allows FC traffic to flow over Ethernet networks. I buy the uniter part, as it clearly positions FCoE as a feasible backbone technology to meld data center networks. But it could very well end up as a case of unite and conquer if FCoE gains sufficient market share as 10Gig Ethernet and Data Center Ethernet take hold. It could even open the door a little wider to InfiniBand, which (as always) is lurking in the wings. In such a scenario, FC could easily become less of an imperative for enterprises.
I've so far managed to stay away from "perfect storm" metaphors, largely because there's nothing perfect about these circumstances. But if all three technologies--iSCSI, SAS and FCoE--continue on their current trajectories, it's bound to have a significant impact on how companies do storage. Toss the cost factor into the mix, and the storm still might not be perfect, but it's certainly beginning to brew.