As published in The Next Platform
It’s elastic! It’s on-demand! It scales dynamically to meet your needs! It streamlines your operations, gives you persistent access to data, and it’s
always, always cheaper. It’s cloud computing, and it’s here to save your enterprise.
And yet, for all the promise of cloud, there are still segments of IT, such as HPC and many categories of big data analytics, that have been resistant
to wholesale outsourcing to public cloud resources. At present cloud computing makes up only 2.4 percent of the HPC market by revenue, and although
Intersect360 Research is forecast its growth at a robust 10.9 per year, that still keeps it well under 5 percent throughout the five-year forecast
period.[1]
If cloud computing is so indisputably awesome, why are some markets stuck under mostly sunny skies? Typically, answers to this question float to the
so-called barriers—security and data movement are the most common scapegoats—but in reality, cloud computing adoption goes beyond finding
the keys to unlock the gates of heaven.
The cold, hard truth is, not everyone pictures cloud computing as Cloud Nine. Some people see the world differently. This shouldn’t be a surprise;
markets have segments. No matter how much you and your friends like them, not everyone else likes Starbucks, Costco, or “Orange is the New Black.”
But even for those with cloud-colored glasses—perhaps especially for those proselytizers, the cloud-noscenti—it’s worth pointing out
that not all of the great truisms of cloud are strictly true.
Lie #1: It’s in “the” cloud.
We talk about the cloud as if it were a singular, magically connected ether-verse, the same way we talk about the web, the internet, or perhaps heaven.
It’s in the cloud, like that’s one place, where your corporate accounting is hanging out with my Instagram pictures when no one’s looking.
This is so obviously untrue that is almost doesn’t bear mentioning, except that it leads to sloppy behavior when users fail to consider where “in the
cloud” is. Who is in charge of the data’s accessibility and stewardship? Who ensures availability of applications? If I want something back from
the cloud, how do I get it, and will it be in the same condition as when I gave it up?
Even among major cloud service providers, services are not always the same, and when “the cloud” is wrapped together with other trends, such as Internet
of Things, our awareness of the cloud is at yet another remove. That instantiation of “the cloud” where your smart lawn sprinkler data is kept
might be a rented partition with Google, or it might be an independent server closet in Alpharetta, Georgia.
Rectifying this lie is mostly a matter of language, along with consumer awareness, and at an enterprise level, it is easily enough managed with service
level agreements. Yet any time we refer to “the” cloud, it belies a certain laziness in specification that sets off alarm bells. Ask yourself if
you are prepared to be asked, “What do you mean we can’t access it? I thought it was in the cloud!”
Lie #2: You only pay for what you use.
The elasticity of the cloud is often touted as its greatest benefit, saving you money through the magic of only paying for what you use. When you need
more, they give you more; it’s a pay-as-you-eat buffet of enterprise infrastructure where they simply weigh your plate at the end of the line.
But what about the food that goes uneaten? Who pays for that? Why, the restaurant does, of course, just as the cloud service provider pays for all
of the unused computing cycles and storage that gets thrown out at the end of the shift.
It still sounds like paying for only what you use, until you consider: The cloud service provider is still making a profit. After paying the power
bill, hiring the administrators, and advertising its services, selling clouds is a profitable business, even if it doesn’t sell all of its capacity.
Who pays for the unused cycles? The buyer still does, built into the price of the cycles that do get bought.
And by the way, there are excess cycles, just like there’s always leftover food on the buffet at the end of the night. Clouds are provisioned for elasticity,
and the cloud provider doesn’t like to get caught without desired capacity.
Lie #3: Cloud computing is definitively cheaper than on-premise.
The notion that cloud computing is categorically cheaper than on-premise draws on two philosophies. One, that there are substantial savings implicit
in only paying for IT on a utility basis—but as we just saw, when IT needs are relatively stable, this is a fallacy. Two, that economies
of scale allow the cloud service provider to acquire and manage IT resources far cheaper than the smaller individual enterprise could.
To be sure, hyperscale companies like Google, Amazon, Microsoft, and Baidu wield considerable buying power, and they even bypass traditional OEMs to
employ ODMs or to purchase components directly from suppliers. But to espouse the volume of their purchases as the driver of savings is to draw
the wrong comparison. Any system vendor buys components in large volumes. The only thing that has changed is the delivery mechanism, to a short-term
rental economy. Depending on your commuting patterns, it may cost you less money to take a bus periodically than to buy a car, but the economies
of scale for the municipal fleet isn’t the fundamental reason.
This isn’t to say that cloud doesn’t make economic sense for buyers. It often does, particularly when demand is highly variable. When an organization
needs a relatively large increase in capability for a relatively short period of time, the pay-as-you-go model still offers a benefit, despite
the implied overhead. Additionally, for many organizations, managing IT isn’t considered to be strategic or otherwise core to operations. In fact,
in some cases, it may be worth a premium to pay someone else to manage it for you.
The Truth of Cloud Computing
There are myriad cloud service providers out there, all competing for your business. Combined with smaller providers, application specialists, and
scientific collaboration sites, there are hundreds of public clouds available. They offer you the ability to outsource any portion of your IT infrastructure
or workload, scaled to the size required, and adjustable over time.
Sure, you have to pay a premium to cover management and overhead. And if you can use the majority of the capacity you buy, it isn’t necessarily any
cheaper to outsource, relative to managing infrastructure internally. But for many users, cloud computing is worth it.
And it has a major role in the market.
Even within HPC, a market that has been resistant to it, we see major growth potential for cloud. In addition to bursting for short-term additions
to peak capability, cloud provides an on-ramp for organizations that are moving into HPC for the first time. There have always been application
consultants to help manage workloads, and now the infrastructure can be outsourced along with the expertise.
If the promise of cloud computing is overblown, it because of the amplification it gets from its loyal converts, enterprises who have found liberation
and agility in outsourcing IT. There will still be segments of the market where cloud computing doesn’t make economic or operational sense, and
even some workloads and data sets for which enterprises reintegrate workloads, bringing them back on-premise out of the cloud.
The truth is, cloud computing is here to stay, and it is growing, specifically because it represents a new paradigm that introduces more capabilities
to more people. But it is not a panacea that will overwhelm the IT landscape. Cloud will not fully displace on-premise, but it will remain an extended
part of the infrastructure—the hybrid cloud, if you will—in most forward-looking enterprises. And it will enable fabulous new applications,
things that we have dreamt of, and even things we have not.
The true leaders in adopting cloud computing are not the organizations that heuristically push everything off-premise based on the faith that life
is better in The Cloud. Rather, the true leaders are the ones who see past the platitudes to determine which clouds to adopt, when and where, and
for what.
With a well thought out strategy, cloud computing can help transform an enterprise, introducing more capabilities, streamlining operations, or delivering
more return for an optimal investment. There are a lot of clouds to choose from, ready to scale to any level of need, with utility pricing plans
that make financial sense for a lot of workloads. And that’s no lie.
Editors Note: Keep an eye open for the free download edition of The State of HPC Cloud: 2016 Edition from Next Platform Press. Available via Amazon
and other booksellers December 20th with a special free download week beginning October 24, sponsored by Nimbix.
[1] Intersect360 Research, “2015 Worldwide HPC Total
Market Model and 2016-2020 Forecast,” September 2016.