In moving to the cloud, many organizations are opting for private cloud computing environments. This model makes a lot of sense as a method to consolidate IT expenditures and resources, automate tasks and introduce new technologies such as virtualization. Often, this model provides a bridge
More on moving to the cloud
A guide to moving applications to the cloud
CIOs aren't letting cloud security risks derail move to the cloud
Is compliance preventing you from moving to the cloud?
Security presents challenges in the public cloud, but you’ve probably already solved many of those challenges in your own data center. Why not use those solutions with centralization and automation to reduce the initial complexity of a cloud project?
Moving to the cloud doesn’t require virtualization
Many enterprise IT shops are moving toward cloud computing while thinking that it’s all about virtualization. In many data centers, a virtualization platform is central to service offerings, but clouds aren’t really defined just by technology. Instead, they involve people, processes, centralization and control. Moving to the cloud promises to consolidate duplicate services within an organization and to automate routine, mindless tasks so that employees are free to work on difficult problems.
A cloud can be a shared infrastructure, which could be virtualized, but it could also encompass physical hardware. Take public cloud services such as Google’s Gmail or Microsoft SkyDrive. Those services don’t use virtualization. Instead, each is built on thousands of physical machines. Some of your private cloud services may be built this same way, especially for services that rely on technologies such as Microsoft Cluster Service, which can be somewhat incompatible with how virtualization environments work.
The goal is centralization, not virtualization. If you can consolidate 50 file-and-print servers spread throughout your organization into three clustered physical hosts, that’s a cloud win, regardless of the lack of virtualization.
Making the transition to the cloud
So how do you move from a virtual environment to the cloud? First, standardize your technology foundations. Virtualization enables organizations to standardize operating system configurations using virtual machine (VM) templates and to automate some of the deployment tasks. It also lets them standardize on areas such as replication, firewalls and other security measures, OSes and storage configurations. Standardize these where possible for both virtualization and cloud computing.
If you cannot create a one-size-fits-all solution, pick several sizes. The goal is to eliminate one-off configurations. Having 10 different types of VMs is much better than having 3,000 individual configurations.
This is also a great time to think about automation, specifically about eliminating repetitive tasks. In moving to the cloud, could you add routinely used software to the virtual machine templates to avoid having to install it? Instead of creating local accounts on each server, would a central Lightweight Directory Access Protocol or Active Directory instance be more useful? Could you begin using configuration management tools such as Puppet or Chef to automatically transform and manage a server’s configuration?
Even a script of routinely run commands helps enormously. Systems administrators take pride in their laziness -- not typing commands more than once if they can help it. Follow their lead and avoid doing the same task more than once without automating it.
Automation is not necessarily about self-service, though. Too often, the cloud is seen as a self-service offering, but IT shops have spent years wrapping process around the act of creating and managing servers for good reason. These processes are often responsible for how a server or service is monitored, how documentation is created or how licensing is handled, for example. Throwing these things aside by offering self-service is a mistake.
Once you’ve laid a good foundation for your cloud with standards and automation, you can begin the more difficult work of surveying your organization for the IT services it runs. Finding the services in use can be challenging. Even tougher is figuring out why each one exists.
People may have good reasons to duplicate services. For example, perhaps the company's main Web servers didn't support a specific technology, so a department created its own. Document these needs and work to extend central offerings to meet them. You absolutely have to be flexible, too. Moving to the cloud unwinds years of work on IT infrastructure in the process of centralizing, and you will find things you never anticipated.
This was first published in July 2012