2011 tech priorities: Private cloud beckons

Before jumping in you need to master virtualization, garner C-level support and have a business case

The indisputable economic benefits of cloud computing for certain applications drive businesses to consider building clouds of their own, but they need to make sure they are prepared before jumping into the cloud.

Consideration of people, governance, price and technology all need to come into play, says Alan Boehme, senior vice president of IT strategy and enterprise architecture for ING Americas, a user of cloud technology and a founder of the Cloud Security Alliance.

5 predictions: How virtual networking will change the network industry

Plus organizations need to climb a virtualization-maturity curve before they have what they need to take on the challenges of a private cloud, says James Staten, a vice president and principal analyst for Forrester Research.

Climbing the curve calls for going through a four-step process consisting of selling virtualization to reluctant users, deploying it in earnest, optimizing the use rates of the server pool, and creating incentives that encourage sharing virtual infrastructure across the business, he says.

During the first stage, which most IT departments have already completed, IT goes begging to find developers within their organizations who will try the technology. In the next stage IT makes it easy to get a virtual machine outside the approval process needed for a physical machine.

Staten calls this the hero phase because IT can reduce the time and cost of adding new resources. During this phase the ratio between the number of physical servers that would be needed with and without virtualization is reduced and becomes a benchmark for success.

In Stage 3, that ratio becomes less relevant as the business tries to maximize utilization rates of server resources across the pool of physical servers, and the key metric is that utilization rate. During this phase, tools and procedures must be put in place to minimize virtual machine sprawl, making sure that virtual machines that aren't being used are shut down so the server resources can be reused, Staten says.

The final stage calls for active marketing of virtualization to divisions of the business and convincing them it's best for the organization as a whole to share resources. The most effective means is to show the dramatic difference in time it takes to authorize and turn on a physical server (a matter of weeks) vs. turning on a new virtual machine in the data center (a day to a week) vs. turning up a VM in the cloud using a standard template (immediately). When a central store of servers can be shared among business units, the business is ready for adopting a private cloud, he says.

For businesses that haven't climbed very high yet on the maturity curve, there are faster ways, Boehme says. For example, a group within IT can be created to set up a relationship with a cloud service provider to carve out a dedicated cloud for the business customer within its service cloud infrastructure. That relieves many of the burdens of creating a business owned and maintained cloud.

But there's more to preparing for private cloud than just virtualization, Boehme says, and training is critical. IT staff needs to break out of their narrow-focused tasks to become more generalists. Rather than having a storage specialist, a broader group should pick up necessary storage skills, for instance. "Let people out of their box," he says. "A lot of people you have already can transition OK if the proper management is in place," to encourage and facilitate the change, he says.

Also, if businesses are hiring while they're making the transition to a private cloud, they should consider younger candidates because they have grown up with virtualization and aren't locked into traditional thinking about how provisioning IT services ought to be done. "They don't know what you can't do," he says.

In addition to developing in-house skills, businesses need to create centralized software development life cycle (SDLC) rules on how the business as a whole builds and tests applications. Existing rules were created to slow the process. "They were put in place to stop a lot of change and make the environment predictable and stable," Boehme says.

With virtualization within cloud architectures, users can provision their own space, with governance aiming more at overarching rules to make sure individuals don't overprovision, so that live migration of virtual machines happens smoothly and that corporate risk policies aren't violated, he says.

IT departments must also establish financial transparency for cloud services they provide in order to get their customers to buy into it. These customers are unfamiliar with consumption-based billing. Traditionally, they pay for physical servers, but with a cloud can be billed per minute, per core, per megabyte, Boehme says.

That cost can vary within the cloud depending on various factors such as what environment the resource is deployed in and time of day. Customers, who are likely also familiar with public cloud services, may not understand why the prices may fluctuate since public cloud charges are always the same. "The agility and flexibility that drive big cost benefits [for the business as a whole] impact the ability to provide understandable information to customers," Boehme says.

Once a business is ready to use a private cloud, it has to pick the right applications to deploy there, says Jessica Carroll, managing director for IT at the U.S. Golf Association. For example, the USGA is planning an app that will support the organization's social media site, but it has no idea how much traffic it will get. That will be hosted in a private cloud that has been carved out of a public cloud provider's network. She would not name the provider but says it was chosen because it was small and its top engineers would be able to respond quickly at any hour of the day, she says.

If the application needs more processing or memory, it can added quickly.

In choosing a provider who'll create a private cloud within their cloud, Carroll recommends onsite visits to witness the features - servers, data center, generators, etc. - that insure the level of service required. "It is really worth the day trip. It makes it all real so you know everything in that contract actually exists because you've seen it," she says.

Find out what the other tech priorities are for 2011.

Editors' Picks
Join the discussion
Be the first to comment on this article. Our Commenting Policies