ISCSI vs. Fibre Channel: The battle resumes

Today and next week we will look ahead to more trends that we are likely to see in the New Year. At this point, it is time to make the obligatory comments about iSCSI and Fibre Channel storage-area networks … and about something else as well.

ISCSI vs. Fibre Channel has been an off-again/off-again battle for years now. Originally, the two were supposed to lock horns in a fight for the corporate IT SAN space. But iSCSI developed slowly, and it became obvious that the battle would come about later rather than sooner. Then it looked like iSCSI would become the “SAN for the everyman,” a less expensive and easier to use format that would work well for smaller businesses and departmental needs and which would leave Fibre Channel products to provide for the high-end.

Now once again the technology sands have shifted beneath our feet and iSCSI – mostly riding the “powers of 10” increments that Ethernet provides (going from 100Mbps to 1Gbps to 10Gbps) – is now challenging Fibre Channel in the area of performance, and typically beating it when it comes to price. Already I see a number of sites running iSCSI SANs at the same sites that house their Fibre Channel ones, and device vendors adding iSCSI connectivity to their high-end offerings. At the same time, market pressure is nudging down the price of Fibre Channel. Increasingly, decisions about SAN technology are being made along financial lines rather than just being driven by what technology is available. So at last the battle is on.

Who will win? That may turn out to be irrelevant, and not just because there is plenty of business out their for both sides of the war. New SAN technology, one that requires no switches, is beginning to appear. As switches often represent a substantial part of the SAN environment, this approach offers the potential to seriously reduce the cost-per-gigabyte of SAN storage.

Small versions of a switch-less SAN are already available in consumer products. In fact, I am running one now. Don’t be put off by that fact that this has such “humble” market beginnings, however. Remember that PCs once started out at the bottom of the computing totem pole, but within a few years grew in power to the point where they were able to knock off all those expensive RISC-based engineering workstations where we once did all our CAD and engineering work.

Remember products from companies with names like Ardent, ComputerVision, Dana, HP, SGI, Stardent, Sun, and my old favorite, Lundy? Perhaps not, but no matter. The point is that once these powerful PCs were the answer to 99% of the market’s needs and once the software vendors caught on to that fact, for most of these vendors it was bye-bye.

If the scalability is what the vendors claim it to be for these switch-less products, and if the performance scales as well, this will prove to be a truly disruptive technology. These are, of course, important “ifs,” but keep an eye out for this. I sure am.

Next week: My comments on blades, SAS and SATA, what’s in store for storage for the small and midsize business market, and a few other things as well.