Implementing a hybrid cloud with automated self-service provisioning of virtualized workloads can help agencies become more agile and collaborative.
We won’t bore you with discussion about how more federal agencies are moving to the cloud. However, for an increasing number of agencies, the question of public vs. private clouds is evolving into a discussion of how to implement a hybrid cloud model characterized by fully automated self-service provisioning of virtualized workloads.
Those on the path to the hybrid cloud must tread carefully, however, because it’s not as simple as just tacking a cloud infrastructure onto an existing environment. That’s especially true in the federal government, where “existing environment” often means using a mixture of legacy as well as next-generation hardware and software.
Keeping that in mind, there are three distinct processes that agency IT professionals should follow to create a foundation for a more automated, agile and collaborative agency.
Review the environment
Before any hybrid cloud deployment takes place, conduct a detailed and all-encompassing review of the entire existing IT infrastructure. This involves a few distinct steps.
First, look for documented, routine processes applied to both physical and virtual servers during the provisioning and decommissioning process that can be automated, either through cloud orchestration or configuration management software. These processes can include manual tasks, running of in-house scripts or simply business processes. Automating these tasks will provide the consistency and scalability essential to the long-term success of self-service hybrid clouds, while maintaining and even improving the agency’s security posture, because automation can mitigate the potential for human error.
Next, establish workload types and overall goals specific to the agency mission, and map these to one or more cloud providers. Many providers use different underlying technologies -- enterprise virtualization, scale-out infrastructure as a service (e.g., OpenStack) and containerized platform-as-a-service -- that are well suited to host specific types of services and workloads.
Then, consider potential integration points outside of the ones directly between the hybrid cloud management software and the individual cloud providers. Hybrid cloud approaches typically deploy an array of hardware and software solutions with varying degrees of overlap between different cloud providers. Full hybrid cloud monitoring and automation requires integration between DNS/directory servers, systems management, configuration management, enterprise storage, load balancers and backup providers.
Now, for the hard part.
Get everyone on board
Hybrid clouds have a tendency to eradicate siloed management approaches, which are all too common in government circles, even in the age of DevOps. Building and maintaining a hybrid cloud requires continual, open collaboration between decision makers and “grassroots” IT operations and developer teams. As such, IT managers would be well advised to identify the key players and sell them on the move toward a hybrid cloud environment by articulating how to tackle the initiative and deliver incremental success.
Start by extolling the virtues of an automated, infrastructure-as-code environment for more innovation, less maintenance and better communication and collaboration. This final point is key, as non-telegraphed changes to the IT infrastructure can easily bring down cloud services. It may no longer be feasible for IT teams in charge of key cloud integration points and services to make significant software configuration changes to data center hardware without coordinating with stakeholders and end users. For many agencies, getting all relevant IT stakeholders in regular cloud integration meetings and in the habit of coordinating and cooperating with each other is frequently accompanied by growing pains. But rest assured, both the short and long-term payoffs are well worth it.
Go one provider at a time
Start with a single cloud provider-- usually the on-premise legacy enterprise virtualization provider -- and build from there, establishing a template for success
Then, strategically add providers. Know the strengths of each and assign the right workloads to the proper solutions. Establish a common template for lifecycle management that covers multiple providers with additions or amendments for each. Use policy, automation and orchestration to tie user workloads to service types. Finally, establish a standardized and documented procedure for reviewing, testing and promoting automation and configuration management.
Finally, it’s important to remember to create the hybrid cloud in a way that’s open and easily adaptable. Maintain an open framework that is built to accept whatever solution -- public, private or combination thereof -- that works best, now and in the future.