Economics of Cloud Storage

In his recent blog, my friend Jan Klincewicz wrote this: "To many, the exchange of CAPEX to OPEX is the real driver behind going to a Cloud model."

Jan, I couldn't have said it better myself. So as it is with cloud compute, so it is with cloud storage. Most folks concerned with running large datacenters (or even small ones!) that have legacy storage arrays are faced with a Hobson's choice: either pay for maintenance to 'keep the lights on' (as a recent CIO I spoke to opined) or take a serious CAPEX hit, downtime risk and much labor (either FTE or contract) to forklift-replace the beast.

So, it's natural that many CIOs would consider cloud storage to do what Jan said, exchange CAPEX to OPEX. In a perfect world, they could merely pay a monthly bill and still meet all their business requirements related to storage - which boil down to these three fundamental aspects of data: store, protect, move.

However, I believe one thing has been forgotten, or at least shoved aside, in this thought process. It is that inhouse storage can be made to be very efficient, very cost-effective, and very simple to operate while at the same time lowering CAPEX and lowering OPEX.

What many CIOs forgot is that an enormous amount of innovation is occurring in the storage field as we speak, and much of that innovation is being translated into products which inhouse datacenters can take advantage of - IF they clear their minds and look around. Remember, Einstein said (paraphrasing) that the definition of insanity is using the same thinking that created the problems in the first place, and expecting different results.

This is not to say that cloud storage should not be considered - quite the opposite. Cloud storage, in particular public cloud, may be entirely appropriate for some data, even for large enterprises. But I believe that an insourced cloud model (what some are calling 'private cloud', but that term is way overused as Jan correctly said) or an insourced refresh of storage with an emphasis on PSAs (performance-starved applications) and the optimization of virtual environments is the optimal way to approach the problem.

So, the question is not "to cloud or not to cloud" - but rather what to cloud and what not to cloud. That is the question. But in that question, do not assume that insourced storage architecture is dead - it is not, and many innovative designs and storage elements exist that compliment a cloud approach.

The economics of cloud storage are in their infancy - and it's going to be very interesting to see it all unfold. So, keep on thinking about your data - because nowadays, it's all about the data. After all, they don't call them servercenters, do they?