The two major trends sweeping the enterprise right now are virtualization and cloud computing. Both build on the same technology sets and offer new levels of flexibility and scalability, not to mention significant cost savings over the traditional ways of doing things.

And yet very little attention is being paid as to how these two elements will work together – or perhaps more importantly, how development and deployment of one will affect the other.

Both technologies have their origins in the early days of IT networking. Virtualization had been a staple of mainframe systems dating back to the early 1970s, while references to cloud computing extend back to the early days of grid computing and the efforts to deploy resources as utilities. In their modern versions, however, virtualization involves partitioning server and network resources to run multiple operating system or application images, while the cloud allows those resources to be distributed dynamically across multiple enterprise infrastructures, either internally or externally. So while the two do not perform exactly the same function, it’s safe to say that a fair amount of interoperability between the two areas is vital for a smooth-running operation. In fact, virtualization is essentially one of the building blocks for a dynamic cloud infrastructure.

“When you’re talking about virtualization and the cloud, without virtualization there is no cloud,” says Shaun Walsh, vice president of corporate marketing at Emulex. “Without virtualization at the server, fabric and storage levels, you can’t do the kind of dynamic allocation that you’re looking for in a cloud.”

Emulex is wading into both technologies through techniques like NPIV (N_Port ID Virtualization) and automated management that allow users to set up new virtual machines, dynamically allocate storage to them, and move them from one physical location to another. Then on the back end, the company is looking for ways to managing virtual SAN technologies according to class of performance, and then dynamically alter switching resources to provide servers with sufficient support, QoS, security and other parameters – all on the fly.

“These are the beginnings of the early cloud computing promise,” Walsh says.

Already, the top virtualization vendors are flocking to the cloud as a means to extend their existing platforms. VMware’s effort is the vCloud initiative. Although still a little vague on details, VMware seems to be looking at ways to bundle virtual applications (VApps) and services across multiple locations through a common set of APIs. The company is in the process of designing the APIs in such a way that they are not tied to a single server or network resources, but can connect requests to virtually any device on the cloud.

Naturally, Microsoft isn’t willing to give up the cloud without a fight. The company recently launched its Azure development platform aimed at fostering cloud-based collaboration and is already lining up its Hyper-V hypervisor and tools like the Virtual Machine Manager to support any cloud-based applications that are created out of Azure.

There is also a wide range of Microsoft APIs that can link the company into any cloud platforms that become available. ADO.net, for example, can be used to link cloud users to a wide range of data sources that are likely to be available through numerous cloud services relatively quickly.

Managing Both

Any discussion on the convergence of virtualization and cloud must center largely on the issues of management and automation. At a time when many IT managers are just getting their heads around the idea of managing virtual servers, adding responsibility for an ever-shifting cloud infrastructure must seem exponentially daunting. But systems experts say there is a way to keep both the virtual environment and internal/external cloud services from spinning out of control. A lot of it has to do with simply getting increasingly disparate software modules in newly integrated management stacks to communicate effectively.

At IBM, the goal is to combine virtualization management and service management tools to ensure both pooling and sharing of resources and delivery of services to users. The trick will be to keep a handle on things as the nature and variety of both virtual elements and cloud services starts to multiply.

“We would expect that in the organizations of the future, you will find highly virtualized infrastructures coupled with service management used for delivery of a myriad of business and IT services over a myriad of architectures, from dedicated workload pools to private cloud solutions to standardized services subscribed through public clouds to federated clouds comprised of potentially dynamic combinations of infrastructures,” says Andrew Wachs, segment manager for virtualization at IBM’s Systems and Technology Group. “Virtualization will enable agile infrastructure around the world to provide more choices for organizations based on service quality requirements, like security and performance, privacy restrictions, compliance requirements and other factors.”

Service automation will also be a key aspect of most future management stacks. CA is moving ahead on this front with a new integrated management regime under the Enterprise IT Management (EITM) solution. A key element is the Data Center Automation Manager, designed to oversee many of the tasks surrounding the provisioning and maintenance of cloud infrastructures. The system provides for real-time resource control, network availability and SLA maintenance – all geared toward dynamically shifting resources to meet peak data demand.

“The need to differentiate services will be paramount for cloud providers to succeed,” says Stephen Elliot, vice president of strategy for CA’s Infrastructure Management and Data Center Automation unit. “In addition, the need to have a profitable business model will require automation to get the most out of IT resources. Finally, the need for cloud providers to respond to and fulfill business demands that have no clear demand patterns will be paramount to their success. Virtualization offers the agility to meet these demands across IT silos.”

CA is already tapped in to VMware environments through its support of the vCloud initiative and the company’s Ready Program, which seeks to foster broad industry interoperability.

Where to Start

These kinds of visions may be all well and good, but IT executives generally have to deal with more immediate concerns – namely, how to get there from here? In that vein, companies like rPath may be able to offer a guiding hand. The company has laid out a five-step process that can act as a template for first building the virtual infrastructure and then the cloud layer on top.

“The three things that are converging are virtualization on both the application and infrastructure levels, grid/utility computing, and the SaaS model,” says Jake Sorofman, vice president of marketing at rPath. “We’re witnessing the transformation of the existing computer model, which used to be very capital-intensive and rigid, to a very elastic model featuring on-demand provisioning of capacity, smooth deployment of applications and dynamic scaling.”

The company’s guidelines spell out a series of investments starting with virtualization of infrastructure and application deployment, followed by experimentation with third-party cloud services like Amazon’s Elastic Computer Cloud (EC2).

“Get your hands wet, your feet dirty,” Sorofman advises. “Deploy some apps. Understand what the cloud looks like and how it operates. Adopt some best practices and use it as a basis for learning.”

The nest step is to start building your own cloud foundation. Develop a reference architecture and codify it as a set of best practices that will allow you to scale the system as needed. Then you can start releasing applications on the cloud to gain mastery over its eventual complexity.

Once you have that architecture in place, you’re ready to connect the full application development pipeline to the cloud for broad-based deployment of those applications. At this point, the final stage is merely to scale up the entire architecture to encompass dynamic workload provisioning across the cloud.

“There’s nothing simple about any of this,” Sorofman says. “It will be fairly complicated and, quite frankly, today’s technologies are not mature enough to support this vision entirely. But the starting point is the general notions of virtualization in the data center and in application deployment.”

At this point in its development, there is still a lot of talk about cloud nirvana. Those who have been around the block a few times no doubt realize that cloud nirvana is likely to be as chimeric as virtual nirvana or even server/storage nirvana.

But the old saying about not letting the ideal get in the way of the very good certainly applies here. However cloud technology shakes out over the next few years, it will most certainly produce a more flexible, dynamic and cost-effective IT environment.