Clouds Cannot Be Contained In A Box

A big part of the value proposition of cloud is to ensure that you have continuous access to your data, and that you've moved beyond the physical limitations of a single box or a single data center or a single geography. While the move to the cloud can allow greater leverage of compute servers and storage, it also provides the ability to move away from aging, monolithic storage and servers, and gives cloud customers access to their data irrespective of any technical issues that may be going on and irrespective of their physical location. Cloud is supposed to be always on, with resources available on-demand, 24 by seven. However, how do you deliver all of this with a cloud that's been imprisoned in a box? Clouds cannot be contained in a box.

In my opinion, and facts are starting to bear this out, you can’t have continuous access to your data if it’s stuck in a single box or a single data center. I’ve been thinking about developing a Trainer Cloud Test. You know, just something simple that can illustrate whether or not you actually have a cloud or are safely securing your data in a cloud. A friend of mine always says the simplest way is the best way. So, let’s keep it simple.

Here’s the Trainer Cloud Test: Want to know if you have a real cloud? Regardless of whether it’s private, public or hybrid, this test will work: If you spill a cup of hot coffee on your cloud infrastructure, will the whole thing shut down?

If the answer is yes, guess what? You don’t have a cloud! Substitute hot coffee with any disaster you like.

Maybe you have a really slick server, or maybe a highly featured new storage array or a rack with both of them plus some networking gear and VMs. But that’s not a cloud. It’s just hardware and software sitting in a single data center.

In my opinion, a cloud needs to be designed to make sure that if someone spills hot coffee on one rack and it shuts down, customers can still access their data and don’t notice a thing. This is possible because data requests can be satisfied from another node in the same cloud. In my opinion, a cloud should have a minimum of two data centers in two completely different locations to actually function like a true cloud and deliver non-stop operation that enterprises demand. Otherwise, you have a single data center with a bunch of really great IT gear in it, and when things start to crash, your customers (and you) are out of business. From my perspective, single data centers don’t make a cloud because they don’t provide continuous data accessibility. One can make 25 copies of information in one data center and claim you have redundancy built in, but that’s still not cloud. If there’s a fire, or another disaster that takes out the data center, your chance of accessing any one of those 25 copies is impossible and nothing but a data pipe dream.

That’s why when I see things like HP’s latest cloud in a box or Oracle’s never-ending cloud in a box products, they make me nervous because of all the confusion they’re causing in the industry. In my opinion, these aren’t clouds, they’re just boxes like they’ve been selling for decades. Granted, they’re faster boxes, but boxes nonetheless.

"Cloud in a box" is an oxymoron. I believe clouds are supposed to be geographically distributed and geographically aware, and not locked in some IT vendor’s box. This means that an EMC V-Max is not a cloud. And neither is a NetApp filer or a FlexPod. And whatever box Unisys is claiming is a cloud today.

What I find interesting is that most of these cloud-in-a-box products don’t even have an object store. They’re just taking existing hierarchical file systems and claiming they are ready for cloud, but it doesn’t work like that. You could pull off this kind of marketing trick in the past, but cloud is disruptive when you look at the technology and what it does to company’s business models. Continuous, geographically dispersed data access is critical for clouds to be real.

An example is Salo, a finance, accounting and human resources staffing services company that moved much of what it does to the cloud, leveraging the security and accessibility that mindShift can deliver to an SMB user. Salo’s CFO, Denise Doll-Kiefer, spearheaded an effort to leverage the talents within her company to select a cloud services and storage provider that delivers the kind of information accessibility across data centers that will allow Salo to remain up and running and to get on about doing its business--and not micro-managing IT.

Another example is the University of Southern California. Just ask the CTO of USC, who is deploying an 8.5-petabyte private cloud from southern California based Nirvanix, with the capacity being split into two different states for continuous data access. "We shifted to the cloud because it provides USC with a geographically diverse and cost-effective way of storing, preserving and distributing our content on a truly global scale," says the CTO.

So, IT vendors, take note: The only real cloud in a box is NASA’s Vehicle Assembly Building (VAB), whose interior volume is so big that it actually has its own weather--including rain clouds.

At the time of this writing, EMC, HP, mindShift, Oracle, NetApp, Nirvanix, Salo, the University of Southern California, the NASA Vehicle Assembly Building management and NASA are not clients of Tom Trainer. Tom Trainer is compensated by Gluster/Red Hat; however, he makes no endorsement of Gluster/Red Hat products nor promotes them in any way in this blog.