Is cheap disk storage really cheap?

The proliferation of huge, cache-laden SATA disks in the consumer market has led to an ever-expanding array of very inexpensive networked storage products for business. More often than not, these devices offer both NAS and iSCSI SAN functions that, until recently, were found only in enterprise-class storage products — at a fraction of the cost. Are these ultracheap alternatives right for you? That depends on who you are and what you do.

A huge range of performance variables separate true enterprise-class storage products from their inexpensive pretenders. The most glaring is transactional performance. Most low-cost storage devices are based on a small number of very large SATA disks rather than larger numbers of SATA or higher-speed SAS/FC disks. These types of configurations will yield extremely anemic transactional performance, which would generally make them poor choices for hosting a busy database or mail server.

[ See Paul Venezia’s InfoWorld review: Small business storage finds a new gear for a more sanguine opinion of low-cost storage. | Looking to revise your storage strategy? See InfoWorld’s iGuide on the Enterprise Data Explosion. ]

What if that’s not what you want, though? If you’re a small business without a whole lot of transactional disk performance needs or a big business that just needs a very large, low-cost parking lot for some big data, are these low-end storage devices a good option?

That question is more difficult to answer than it seems. The answer depends less on the actual technology at play and more on who is going to be supporting and depending upon it.

What “enterprise class” really means

First off, let’s take a look at what most people mean when they talk about an “enterprise class” storage array. Higher price is certainly part of it. But to my mind, the real definition has everything to do with how the original manufacturer designed the product to be used — keeping in mind that this is not always synonymous with how it is being marketed (a lesson that can be expensive to learn the hard way).

An easy test is to imagine yourself using the product for whatever you want to use it for — maybe running a virtualization cluster for a small business — and then having the device utterly fail. Are you relying on the storage vendor to resolve the situation for you, or do you have your own recovery plan to throw into action? When you call the storage vendor to get support, what will the reaction be? How deeply will your users be impacted? Can they still do their jobs?

Chances are, if you’re running an enterprise-class storage product, the “failure” of the product wouldn’t be a failure at all. It would be an amber light on a controller, power supply, fan tray, or disk and a warning telling you that you should probably get it replaced. You call the vendor (assuming you bought support) and they ship you a new one. You slide the replacement component into the device and off you go. Production hasn’t been impacted and you’re back where you started.

On the low end, be prepared

If you’re running a low-end device, you might not have a support number to call. Even if you do, don’t plan on getting a replacement component the same day or even the same week. Worse still, don’t imagine that the component you get is going to simply fix the problem all on its own. In many cases, a hardware failure in a low-end NAS/SAN device will necessitate a full restore of the data that was on the device onto a completely new device. Most of the devices in this space are not designed to be modular beyond having RAID and replaceable disks. If you have a controller failure, you’re probably starting over with a backup.

The bottom line here is that if you have the time and availability to jump in and replace the device (perhaps with a spare that you bought because you were saving so much money), a low-end unit might work for you.

Gauging the cost of downtime

The other side of the equation is who is relying on the data the device is storing. If you’re a larger business using the device to store your workstation deployment images, you can probably get away without having it around for a day or two while you get a new device and restore the data. If you’re a relatively small business and are using the device to provide the storage for all of your servers (virtualized or not), you really need to look at how much the absence of that data will cost you.

For example, imagine a small business that needs shared storage for a virtualization implementation. Let’s say the company in question is a local insurance agency with 60 employees. They underwrite with a bunch of different insurance providers and have a document imaging system, so they have more servers than the average small business, which makes virtualization attractive. However, the relatively low number of users won’t put a big transactional load on the storage, so a lower-end storage device works from a performance standpoint. The question becomes, do they spend $2,500 (or even less) on a lower-end NAS/SAN solution, or upwards of $20,000 on an enterprise-class SAN solution?

Well, who is supporting it? If a dedicated IT resource is on the payroll — a somewhat unusual occurrence in small businesses — that person may be well-suited to recover from a storage failure. Or not. If not, there’s a good chance they will pick up the phone and call in some kind of outside contract help to diagnose the problem and work out a solution. Assuming that process takes a day or two, that cost could easily dwarf the cost of the storage device that just failed.

And who is depending on the storage? In our example, our 60 insurance agents and support staff suddenly have no access to most of the information they need to do their jobs. If a customer calls, they may have no idea what policies the customer has or how to resolve claims. In this sort of business, productivity will plummet to perhaps 10 to 20 percent of a normal day and cause the business significant embarrassment with customers. Suddenly, it starts to look as if the prospect of a one- or two-day outage might justify the cost of an enterprise device.

So be careful when you venture into the land of low-cost storage. Take a hard look at what your storage will be used for — and how it’s going to get fixed when it breaks — before you congratulate yourself for saving a dump truck full of money. Sometimes, the dump truck you know is better than the one you never saw coming until it ran you over.

This story, “Is cheap disk storage really cheap?,” was originally published at InfoWorld.com. Read more of Matt Prigge’s Information Overload blog and follow the latest developments in storage at InfoWorld.com.