it en



Use the hybrid cloud model to unite disparate cloud computing visions

IntroCloud projects often create so many questions that enterprises don't know exactly where to start. There's the public versus private cloud debate, the decision over which cloud service architecture to adopt and even the argument over who drives the cloud -- individual business units or the IT department. No matter how you frame it, enterprises need to think about whether to approach the cloud using an "outside in" or an "inside out" approach. The outside-in view of a cloud model Line of business (LOB) planners think of cloud in the "outside in" model; they want cloud computing to make IT a more tactical component of operations. Those of the "outside in" cloud camp see the need to purchase capital equipment and software, sustain a datacenter and support internal IT activities as a potential cost issue and a definite impediment to agility. These planners would like to visualize all of IT as Software as a Service (SaaS), with the ability to quickly commission and decommission applications in response to business needs. This is the "outside" vision of cloud planning -- start with how you want the cloud to look to end users. Looking at cloud from the inside out On the other hand, many organizations take an "inside out" approach to cloud, viewing it as an alternative platform for running applications or as an extension of the datacenter. This puts the emphasis on securing public cloud services that match the technical structure of their current datacenter, allowing applications to run seamlessly when the datacenter is overloaded or fails. "Inside out" thinkers consider cloud services as either Infrastructure as a Service (IaaS) or Platform as a Service (PaaS), depending on whether there is a single dominant software platform on-premises or whether it supports multiple operating systems and middleware. This vision of cloud computing starts with what the enterprise has and evolves it. Cloud computing from all sides The key to a successful cloud deployment is to combine both sides -- the "outside in" team with the "inside out" team. And one way to create harmony between both camps is to make all enterprise applications look like SaaS. Doing so is largely a matter of customizing end users' views by drawing data from available APIs into their graphical user interfaces (GUIs). In most cases, this process is independent of the cloud, but it's important to ensure that the orchestration tools used to accomplish this can access cloud-hosted APIs. All SaaS applications operate through APIs; if a business hosts its applications in the cloud, these apps must also work with cloud APIs. Cloud service providers can offer guidance on how these cloud app APIs can be accessed, and application integration tools are also available to help with this task. Flexibility at the resource level is another concept that can bridge both approaches to cloud computing. Creating an elastic pool of application resources that contains both on-premises servers and public cloud services will typically be the key to uniting "inside" and "outside" cloud models. The challenge is to seamlessly integrate internal IT resources and cloud resources, which admins can approach either as building a private cloud or extending existing virtualization tools. Building a hybrid cloud to bridge cloud models Some companies implement a cloud strategy that includes public and private cloud capabilities. Some popular cloud platforms, including OpenStack, CloudStack, Eucalyptus and Hadoop, allow companies to mix and match public and private cloud resources. For these mix-and-match cloud options, it's important to conduct a pilot test; any specific interactions between resources in the public and private cloud must be spelled out in the service-level agreement (SLA) or contract with all cloud service providers involved. Additionally, many enterprises build hybrid clouds to reap the benefits of both public and private cloud. However, to build a hybrid cloud, IT departments must first ensure they've built a properly functioning private cloud. To do so, it's important to create a resource pool without private cloud software; the technical challenge is in the load-balancing process. Infrastructure as a Service (IaaS) and Platform as a Service (PaaS) can provide resources on demand that, to end users, function no differently than those in a virtualized datacenter. To bridge this private cloud with the public cloud, look for tools that offer server load-balancing capabilities and are compatible with virtualization software being used. At the network level, creating a flexible hybrid cloud means selecting network tools that can be virtualized in the same way you would virtualize servers and storage. Virtual networks (vLANs) and virtual private networks (VPNs) may be the best framework for creating flexible clouds, but connecting public cloud resources with a VPN requires special support, often from network vendors and the public cloud provider.
<< Back