This is the third installment of a multi-part series exploring why our IT organization is so aggressively transforming EMC’s own datacenters into Private Clouds. In earlier posts, I described how our IT strategy shift began, and what gave our senior IT folks a newfound sense of urgency.
In this post, we look at how EMC IT decided how to handle the “cloud issue.”
Clearing The Clouds
There’s a simple reason why cloud-computing models are so attractive. They offer businesses a tantalizing promise: transforming IT from a fixed cost to a variable cost. Does this sound like same promise outsourcing offers? It should. When the outsourcing wave began several years ago, company execs began saying to their IT departments, “you need to offer the same benefits as these outsourcing companies, or you’ll be replaced by one.”
It’s no surprise, then, to see this scene being replayed with a script that replaces “outsourcing” with “cloud.” After all, most vendors pitch cloud as effectively outsourcing an entire IT stack. Some go as far as saying it explicitly: “The corporate datacenter is dead. Everything will exist in the cloud.”
There’s no denying the powerful attraction in such a message. But there are some implications that bear examining. First is that “cloud” is supposed to mean “external.” In fact, let’s be clear on this: it’s intended to mean whatever vendor is currently making the pitch. It also means transferring a business’ control over IT service levels, availability, and security to that vendor.
Additionally, these public cloud models assume customers will replace their apps, or rewrite them to make use of a vendor’s external resources. Choosing a vendor’s APIs makes it “sticky.” Good for the vendor, bad for customer choice. No problem, there are cloud standards, right? Oh, wait…
Meanwhile, a bunch of us marketing and engineering types at EMC were working out how we could offer the benefits of cloud computing—flexibility, efficiency, and dynamic, on-demand resources—while keeping control and choice in customers’ hands. One answer presented itself right away: virtualization. After all, it exists for the same reasons.
It soon became clear, however, that virtualization alone would not be enough. But it could form a foundation for enabling an open marketplace. And that marketplace will drive the economics from fixed- to variable-cost models.
Think this sounds too good to be true? It’s happened before. Another form of industrial infrastructure went through just such a transformation: transportation and shipping.
The big innovation in the shipping industry was the adoption of a standardized, multi-modal shipping container. Businesses no longer needed to own their own fleets of trucks, trains and boats, nor did they have to face the costs and complexity of converting shipments from one mode of transportation to another—never mind from one vendor’s cargo carrier to another. They can just put their stuff in a standard container, and it goes where it’s supposed to, regardless the type or number of carriers involved.
What does that have to do with IT? Well, imagine a company datacenter with its workloads in virtualized containers. Most discussions of moving these containers around assume they stay inside the corporate firewall. The “aha!” moment for us was realizing the power of lifting that restriction.
Imagine an open marketplace of external service providers, using the same containers you’ve standardized on, competing for your business. Imagine having the option of moving your applications where—and when—you see fit. Imagine being able to move your containers among a flexible mix of internal and external sites. Imagine being able to move on little or no notice. Imagine paying only for resources or capabilities you actually use.
This is the central idea behind Private Clouds. We chose the word “Private” because choice—and control—is in the hands of the customer. A Private Cloud can include both internal and external resources as long as you, the customer, can still exert control over your containers. We use the word “public” when it’s the external vendor that’s in control.
From our point of view, this is much better than the typical “re-write and outsource everything to me and only me” cloud model. Yes, next-generation SOA based applications can flourish in such a world, but so can 30-year old COBOL apps. Best of all, it leverages a technology that’s already proven itself, and has become mainstream in business IT. And that means business IT has a smooth migration path available instead of a disruptive “rip-and-replace” proposition.
When EMC IT learned about this vision for Private Clouds, the reaction was similar to that of many customers we’ve worked with: a mix of relief, excitement, and questions. There are operational and economic implications that would still need to be worked out, but it was clear to EMC IT that it was already on the right path, and could proceed with its plans with some minor course corrections.
But EMC IT also did something we didn’t expect. I’ll describe that in my next post!