A Cloud Rosetta Stone?


Tom Gillis, Contributor

January 16, 2015

The tech industry is incredibly innovative—in technology. But I am always stunned by the industry’s periodic lack of imagination when it comes to tech marketing. Here’s what usually happens: Someone introduces a fundamentally new idea and coins a phrase around it. Then everyone else just piles on and muddies the water, to the point where you need a Rosetta Stone or a decoder ring to be able to work out what is really being said.

A great recent example of this phenomenon is “software-defined networking”—a new and important idea introduced about five years ago. After VMware validated the space with its frothy $1.3B acquisition of Nicira, “software-defined” mania swept the industry, to the point where a very credible executive from a reputable tech incumbent stood up at a recent conference I attended and introduced the world’s first “software-defined hardware” and held up a new blade server. It felt like the Iraqi defense minister insisting that Iraq was winning the Gulf War when you could see U.S. tanks approaching in the background of the live TV shot. Sheesh.

Perhaps the greatest marketing confusiality of all surrounds the use of the word “cloud.” It has reached the point that the double-speak around the word has become so pervasive it even has its own name—“cloudwashing.” In 2012 Gartner’s David Mitchell Smith, one of the most respected industry analysts, portrayed cloudwashing as a “term describing how vendors (and IT) paint the cool cloud term on whatever they have, regardless of how little cloudiness their offerings exhibit.”

Bad as it was in 2012, the confusion has only progressed. “Private cloud” is one of my favorites. This term is now so overused that it seems some vendors would have you believe a private cloud is created whenever a few servers are put into a rack. Fact: A private cloud is still a data center—indeed, it’s a very complex data center that requires a lot of sophisticated engineering. Calling one of these large-scale engineering projects a “private cloud” is like calling an oxygen tank “private air.” The name seems purposely designed to confuse.

That said, at least the use of the term “private” lets you know that you are dealing with something different from the “public” cloud. The same can’t be said for possibly the most egregious use of cloudwashing—the practice of taking dedicated hardware, putting it into a colocation facility, and calling the resulting service a “cloud.” Running dedicated hardware in a remote colo is NOT a cloud—it does not bring the efficiency and elasticity to bear that customers are looking for from a cloud solution. It’s basically a re-spun version of managed hosting—an offering that has been on the market for many years and has limited appeal. The underlying architecture typically relies on the same costly, dedicated hardware systems that power an enterprise data center. It has all the limitations of a dedicated hardware deployment, but it’s even more expensive and cumbersome because another vendor is in the mix.

The really interesting cloud is the hyperscale public cloud, delivered by the likes of Amazon and Google. These providers have built a totally new data center architecture with new building blocks. Instead of the highly engineered “scale-up” systems of yore, they have a near-infinite array of identical “scale-out” processors. This new architecture is much lower in cost and incredibly flexible—you pay only for what you use. It is revolutionizing the data center.

But a trade-off occurs with the hyperscale public cloud. To achieve this massive scale, resources need to be shared with other tenants. And as anyone who has sat in rush hour traffic on the freeway knows, too many tenants can sometimes cause problems. Multi-tenancy creates resource contention and therefore performance fluctuations. As a result, achieving predictable, highly available performance on the public cloud can be challenging. Lydia Leong of Gartner recently wrote, “When people talk about performance consistency, they’re generally referring to storage and network performance. … Predictable storage performance is a very difficult engineering problem.”

Unfortunately, this challenge leads us back into the world of cloudwashing. The easy answer to predictable performance is all too often to simply throw dedicated hardware at the problem. Many traditional hosting providers are taking ultra-expensive, traditional enterprise-grade storage systems and standing them up in colos and calling the result an “enterprise cloud.” Even the best, most efficient cloud providers are solving the storage problems by using “provisioned” I/O, or a dedicated chunk of a physical storage system. This arrangement is also cloudwashing—although it works, it doesn’t offer elasticity, or the ability to change on the fly. And so it goes—even true cloud vendors are diluting the term for the sake of marketing, making it difficult to distinguish a cloudwashed cloud from the real, innovative, groundbreaking kind of cloud.

The good news is that genuinely innovative technology typically wins in the end—it delivers benefits that can’t be matched by pale imitations. So despite all the hoopla, we will eventually end up with a cloud that is truly the promised perfect storm—low cost, highly elastic, and easy to plug in to. Until then, keep your decoder ring with you at all times or you might get washed away!

This article was written by Tom Gillis from Forbes and was legally licensed through the NewsCred publisher network.

Great ! Thanks for your subscription !

You will soon receive the first Content Loop Newsletter