With Microsoft and Google both heavily utilizing containerized data centers for compute capacity, it’s hard not to think about the applicability for enterprise and government private clouds. While containers vary widely, the general concept is that they are a self-contained data center environment complete with network, compute and storage, as well as necessary cooling.
At massive scale, such as Google, the container itself becomes the replaceable component rather than focusing on the individual components. Each container is a small subset of the total compute required for search and will not cause a noticeable service outage if it fails. This provides a standardized way to handle compute on massive global scale.
For the average enterprise or government customer, container systems will not work in the same fashion. This is due to the scale of any given application. In these cases, if a container fails, multiple applications and services will be affected. This means that additional considerations must be factored in to ensure that the container will have sufficient up-time. Things like redundant power and cooling are key considerations at private cloud scale. With the built-in density and compact space of containers, any brief interruption in cooling can quickly lead to component failure.
While containers have unique challenges, they can also lend themselves well to private cloud environments because of the flexibility provided by automation and orchestration. The fluid provisioning provided by private cloud tools can allow containers to be added on demand or incorporated with existing data center infrastructure. Additionally, containerized solutions can provide a cost effective means for disaster recovery and business continuance purposes.
Various sizes exist for containers, from single environmentally sealed racks to full cargo shipping containers. Additionally, containers range in levels of environmental support and customization. For
example, specialized containers are available armored from small arms fire and equipment mounted on military grade shocks for rapid deployment in combat zones. While this is not an everyday use case, it speaks to the flexibility of container solutions.
Overlay a private cloud software platform on a containerized infrastructure, and you’re basically provided with ready-to-go compute power that can be deployed on demand in many different environments. Depending on the scale of your environment and the size of the container being used, containers can be used in a fashion similar to building compute pods in a traditional data center. When additional capacity is required, another container can be added and plugged into the pool of compute resources.
Private cloud automation and orchestration is the key to bringing out the value of these containerized solutions. Software provisioning of the underlying hardware means that there is less need for onsite staff, except in failure scenarios, allowing greater flexibility in container placement. Additionally, the resources can quickly be plugged into platform as a service (PaaS) or infrastructure as a service (IaaS) solutions, running either in traditional data center space or other containers.
The ability to place thousands of virtual machines in a single 20-foot shipping container complete with the necessary environmentals makes for a compelling alternative to traditional brick and mortar data centers, which are typically wrought with inefficiencies. Containers can be deployed without the costs of facilities or the time required to build them, and can be used to replace or supplement traditional data center capacity.