How the Data Center Will Grow Up in Three Years
While Docker containers clearly have the potential to transform the way organizations think about virtualization, IT organizations still require an operating system framework through which they can deploy and manage those containers. Today, Red Hat is making available a public beta of one option for accomplishing that goal in the form of Red Hat Enterprise Linux 7 Atomic Host.
Mark Coggin, senior director, product marketing for the Platforms Business Unit at Red Hat, says Red Hat Enterprise Linux (RHEL) 7 Atomic Host is a lighter weight operating environment specifically designed to run development and deployment tools, hosted application services for messaging and integration, container management and orchestration based on the Kubernetes project led by Google, and the underlying container host infrastructure. It can optionally be deployed directly on a physical server or as a guest running on top of a virtual machine. In addition, Red Hat is including SELinux support to make it easier to more securely run containers in isolation.
Coggin says there will clearly be a class of customers that want to deploy containers on an operating environment that runs directly on top of a physical server. But depending on the attributes of the applications, Coggin says other organizations may still want to run containers in isolation on top of virtual machines that they may eventually opt to deploy directly on top of a more traditional operating system environment. They may also choose to deploy containers on top of a platform-as-a-service (PaaS) environment such as Red Hat OpenShift. It’s too early to say which of those deployment environments will become the dominant vehicle for deploying containers. But it is clear that Red Hat is betting on all three.
Coggin contends that because containers are easier to spin up than virtual machines, they tend to lend themselves better to applications that need to quickly access additional compute resources for a short period of time, while virtual machines may lend themselves better to applications that run more predictably over a long period of time.
Whatever the deployment model chosen, IT organizations are clearly soon going to have to come to terms with managing containers alongside virtual machines. The decision they face now is to either try to accomplish that using their existing management tools or acquire new ones that have been specifically designed for managing containers.
