The three routes to cloud computing's future

Ten years after the creation of Salesforce.com, the future of cloud computing is not in doubt; it is just being heavily debated. Two opposing views of how cloud computing will play out--especially enterprise cloud computing--are making the rounds among thought leaders and customer decision makers alike. Interestingly, there is enough to question about both approaches that a third option may, in fact, gain importance.

What all sides agree on, however, is that some form of cloud computing is coming your way. As always, the devil is in the details.

Marc Benioff, Salesforce.com's "pull no punches" supreme leader, represents one of the debate's extremes. At "Whose Cloud is it Anyway?"--a cloud-computing roundtable put on by TechCrunch recently--Benioff stated (the emphasis is mine):

(Microsoft was) a company that...had a lock on the entire industry in terms of innovation, and was able to hold it through a monopoly. So, that is really broken down through a new, next generation paradigm, which is cloud computing; which is no software, no hardware, don't hire anyone, just sign up to these various cloud platforms and pick the flavor that is appropriate for your application.

In other words, it's not cloud computing to Benioff unless the IT department doesn't have to directly handle any form of technology beyond a browser or perhaps an SSH terminal application. This is the very definition one would expect from the leader of possibly the world's biggest software-as-a-service provider.

It is a call to jettison traditional IT altogether, and focus efforts on leveraging the work of professional providers of IT applications, platforms, infrastructure, and services. By this definition, it is indeed a complete change in IT paradigm.

This view is echoed by the current Wikipedia page for cloud computing, as originally authored by Sam Johnston:

Cloud computing is Internet ("cloud") based development and use of computer technology ("computing"). It is a style of computing in which dynamically scalable and often virtualised resources are provided as a service over the Internet. Users need not have knowledge of, expertise in, or control over the technology infrastructure "in the cloud" that supports them.

The 'internal cloud'
At the other end of the spectrum are those who believe the road to cloud computing begins at home. The starting point for any enterprise with existing IT infrastructure investment, according to this camp, is an "internal cloud." An internal cloud applies the concepts of cloud computing (on-demand resources, pay-as-you-go pricing, and the appearance of infinite scalability) to resources wholly owned by the enterprise consuming the service.

There is no doubt that it is a view expressed by much of the traditional IT industry, but there are other voices out there as well pointing out the value of providing multitenant, on-demand, at-scale architectures to internal customers. Internal clouds are appealing to IT departments at many levels, though obviously they are not going to provide the economies of scale that public clouds will offer over time. (For a really good explanation of why large public clouds will dominate the next generation of IT, see the University of California at Berkeley paper titled "Above the Clouds: A Berkeley View of Cloud Computing".)

The strength of the "own nothing" argument is difficult to miss. Benioff put it very well. Don't spend money up front on things that aren't core to your business. Get them as "on-demand" services, instead, and pay for them only as you consume them.

The benefits of internal clouds, however, are a little more subtle. Most proponents will point to the inability of most public clouds to support legacy applications, while internal clouds can be built to handle old and new applications alike. Perhaps the most pervasive argument, however, is that internal clouds allow you to maintain control over security, service levels and regulatory compliance in a way that public clouds are not yet able to offer.

So, what is an enterprise to do? Choosing an "own nothing" approach, like any other paradigm shift, is extremely disruptive and requires a major overhaul or outright replacement of existing IT software assets.

On the other hand, choosing an "internal cloud" approach really doesn't gain the full benefits of public cloud computing offerings. With much smaller scale, the economics are not in internal cloud's favor. As this year and the next progress, I would expect to see it less and less justifiable to rely solely on an internal cloud.

The 'private cloud'
The term "private cloud" is becoming associated with a third option--an option that has fundamental implications to the way in which enterprise customers will approach cloud computing:

A private cloud consists of IT resources under the control of the enterprise consuming it. Those resources may be owned by the enterprise, consumed from a public cloud provider, or some combination of the two. The only requirement is that the resources be under the direct control of the customer under a unified management system, as opposed to each separately consumed offering being individually managed through the interfaces provided by their respective owners.

Many of you may be thinking "hey, that's just the definition of a hybrid cloud", but there is an important, though subtle distinction to understand.

• A hybrid cloud is the use of both public and internal cloud capabilities to meet the needs of an application system.
• A private cloud meets the needs of an application system by any combination of public and internal cloud resources--and that combination can change moment by moment.

Private clouds, by this definition, overcome the "rewrite everything" effect of "own nothing" cloud computing. On the other hand, they provide the degree of trust that enterprises were seeking from internal clouds, including the ability to change the mix of cloud services consumed completely at their own discretion.

In the end, I think the debate will evolve away from "own nothing" vs. "internal clouds", with the latter being replaced by "private clouds." Then, over time, supporters of the "own nothing" vision will come to realize that private clouds give them a direct route to migrating all application workloads from wholly owned infrastructure to public clouds, achieving their vision.

Meanwhile, the enterprise continues to operate with the perception that everything is running in their own data centers, under their complete control. In the end, I think that is the factor that will make private clouds the winning enterprise cloud computing model in the years to come.

So, which is it for you? Will you be taking Benioff's advice and cease to directly purchase software and hardware? Will you play it conservative and insist on turning your own resources into a cloud before venturing out in force to the public cloud?

Will you leverage both approaches as makes sense, a la David Linthicum's frequent advice? Will you pushing the boundaries of what you call your IT resources to include third party services, yet tie it all together within one "trust boundary"? Where do you fall in the great cloud computing debate 10 years after the creation of one of its bellwethers, Salesforce.com?

See also:
• Salesforce.com: Pondering the next 10 years
• Cloud computing: How we got here

You can follow James Urquhart on Twitter.

[KC Cowan]

The distinction between the Wikipedia cloud, internal cloud, and private cloud all make sense. They are all variations on how to provide infrastructure upon which I'd run my applications. OTOH, the way I read the Benioff quote, it speaks to providing turn-key applications directly. We generally call it SaaS, and I don't think we help the conversation by mixing it up with a discussion of infrastructure.

In summary, I view "cloud computing" and SaaS as distinctly different, but related, concepts.

KC

[jacomo]

My question is how would you define a Telco/Service Provider creating a Cloud Computing based data center that would provide both Local Subscribers direct local access to centrally located App/Servers via its IP Core Network and connected to and reselling major Nationwide Cloud Computing/SaaS Services as well.
Approach would have the Local SP connect to the select Natiowide CC/SaaS sources via a high speed data link (IP) and seek both Processing and Apps services as well as a Dual Home Tier#1 Internet link.
This would maximize susbcribers access to and use of the various Content/Apps and Internet links as well as its local Telco/SP.

Jim A.

[KC Cowan]

RE: The Telco scenario ...

For the case of local access to app/severs, it depends on whether you mean applications, such as e-mail, or servers that could run any application, such as hosting my virtual machine for me. The Telco providing e-mail is a SaaS offering. The case of the Telco providing a server is either a managed service provider (circa 2005) or a private cloud.

Reselling a SaaS service is still a SaaS service.

Just my 2 cents
KC

[fdunn3]

CLOUD this, CLOUD that. You would think someone just invented a time travel machine.

Get over it. It is not going to be a major factor anytime soon, at least securely.

[mikedatl]

I think that was the whole point of this post. If you want cloud but want it securely then do it internally right now. If you don't care about the security but want scale or reach or DR then do it externally.

Interestingly enough I've seen a lot of specialize clouds being built that cater to a specific industry. There's Pharma outsourcers building FDA approved clouds for the Pharma industry. You've got financial companies building PCI compliant clouds for credit processors. Etc.