High-End Disk as Backup Target?

Using high-end disk array equipped with low-cost high-density Fibre Channel drives as a backup target, as my colleague Tony Asaro recently put it, would be like ordering a Big Mac at Ruth's Chris and paying $50 for it. Nonetheless, it is certainly an idea worth pondering. After all, the idea of using a midrange disk array for backup purposes was just as (or even more so) foreign a few years ago. But the advent of cheap ATA/SATA disk drives quickly changed the mindset of end-users. So, why couldn't the availability of low-cost high-density Fibre Channel drives similarly change, or broaden, the role of high-end disk arrays?

What do I think? Well, my gut feeling is that organizations will be very hesitant to leverage costly, specialized high-end disk arrays for backup purposes. Now, that doesn't mean I think no one will, but I don't think the response will be great enough to define a whole new market for this class of system.

Organizations may temporarily use these arrays as backup targets, but my guess is they'll continue to leverage other lower-end disk products or VTL for backup purposes. It just makes economical sense. I argue that it's the cost of the software that runs on these platforms not the hard drives themselves that make using high-end disk arrays as backup targets cost-prohibitive.

A  more likely scenario would be for organizations to leverage low-cost Fibre Channel drives to build tiered storage environments, matching drive types to the value of the data that is being written. The more critical the data, the higher-performing drive used. The less-critical, the lowering-performing drive used. This makes perfect business sense.

As for the argument that leveraging high-end arrays for backup addresses (potential) scaling issues with currently available disk-based backup targets, including VTL, I argue that this is a non-issue right now. VTL products from vendors such as Diligent, EMC, HP, and Sepaton, etc., scale into the petabyte range, and data de-duplication technology (available from Diligent and Sepaton) further extends the "scale" of these products by only storing "unique" data. Clustering archictects (such as Sepaton's) also allow end-users to easily scale their VTL environments according to data requirements. Bottom line: I have yet to run into an end-user who has voiced any concerns about the scalability of existing disk-based products.

So, Mr. and Mrs. End-User, what do you think? Would you use your high-end disk arrays for backup? I'd be interested to hear what you have to say.