Practical Analysis: The Storage Stockholm Syndrome

Our 2010 State of Enterprise Storage Survey had one major surprise in it. For the first time in the survey's history, meeting storage needs eclipsed data security as the top concern among storage architects--meaning they don't have the budget to keep up with demand.

That got me thinking: What's the right price for a terabyte of enterprise SAN storage space? Consumers now pay less than $100 for that much storage on one spindle, so that's the low-water mark. Businesses obviously pay a premium over that rock-bottom price for systems that are better engineered and have lots of software and hardware features that make them "enterprise ready." But that premium is now 30 to 200 times, depending on the performance and features.

There's probably another industry with this kind of pricing variance, but I can't think of one. What if that were the case in the auto industry, with Joe Consumer paying $20,000 for a car while the fleet manager for his employer forks over $4 million for a tricked-out version of the same five-seater? Hard to imagine.

While I pondered the plight of enterprise storage architects, the CEO of Coraid stopped by my office. The company has one of those products that might not make it into the mainstream but probably should. It has developed a protocol it calls ATA over Ethernet (AoE), which lets you do away with protocols like TCP/IP, iSCSI, and FCoE and use plain old Ethernet to transport storage data. Ethernet provides checksums and in-order packet delivery, while the Coraid initiators and terminators handle any packet drops. Coraid's premium for enterprise-class storage is five to 10 times the consumer prices for drives. And when it comes to SAN networking, forget the $500-per-port Fibre Channel director-class switch; a 24-port or 48-port Gigabit Ethernet switch costing $500 to $1,000 will do the job just fine.

So what's the right price? Is Coraid pulling a fast one, or has enterprise IT been held captive by storage vendors for so long that we just don't know any better? For a good many applications, it's the latter. Particularly in Linux and VMware environments, you can build systems that perform just as well and have about the same features as high-priced Fibre Channel or iSCSI systems. At least that's what the National Institutes of Health thinks; NIH is using the Coraid system to store human genome data.

But this isn't just about a startup making an interesting product and landing a marquee customer. For the fourth quarter of last year, EMC reported that its earnings per share rose 54% compared with the year-earlier quarter, with the vast majority of that income coming from its high-end storage systems. No one begrudges EMC its profits, but it seems that a little disruptive technology might be worth a look here.

Storage architects are a conservative lot, and with good reason. You only need to lose the company's data once to find yourself looking for a new gig, and no one ever got fired for buying ... EMC, IBM, HP, HDS, and so on. But maybe it's time to take some managed risks and evaluate some innovative--and significantly less expensive--options.

Storage industry heavyweights should also be more aggressive about adopting technology like lossless Ethernet, and in decoupling software pricing from storage volume size. Bleeding your customers dry rarely works as a long-term strategy.

Art Wittmann is director of InformationWeek Analytics, a portfolio of decision-support tools and analyst reports. Write to him at [email protected].

To find out more about Art Wittmann, please visit his page.

Register to see all reports at InformationWeek Analytics.