Private Cloud Automation, Orchestration, And Measured Service

Many infrastructures are labeled "private cloud architectures" when they are really just a subset of a private cloud. The key points that tend to be missed, based on the NIST definition, are on-demand, self-service and measured service.

Joe Onisick

June 23, 2011

3 Min Read
Network Computing logo

In an IT world where everything seems to be getting cloud-washed, private cloud is an area that may be experiencing it the most. Many infrastructures or infrastructure stacks are labeled "private cloud architectures" when they really just comprise a subset of what a private cloud is. The key points that tend to be missed, based on the NIST definition, are on-demand, self-service and measured service.

Many infrastructures described as private clouds are really talking about some subset of private cloud components, which I have referred to as converged infrastructure. Typically, it consists of some level of consolidation, virtualization and integrated hardware, possibly with some automation, monitoring and orchestration. By definition, a true private cloud includes automation, orchestration and service monitoring.

Automation is the foundation on which orchestration gets built. It is programmed processes in place for repetitive tasks. An example of a process that is easily automated is server provisioning.

In a non-automated world, an IT consumer, such as a developer, requests a new server or set of servers. The request will require network, storage, compute and security provisioning, etc. The developer may then have to seek specific approvals, ensure completion, and so on. While there are many steps to this process, they are very static in the way in which they are carried out, even though they are currently manual steps. These steps can be mapped out into a workflow, and automation or scripting tools can be fairly easily put in place to take away the manual touch points of this process. That is automation. Network operation center (NOC) administrators can input the required unique data and kick off a series of scripted processes that will provision the various data center resources once approvals are in place.

Automation simplifies processes and alleviates repetition in static workflows, but it is not orchestration. And without orchestration, we’re still not talking cloud. Orchestration involves tying disparate automated processes and IT resources together using workflows, and typically providing a portal from which those workflows can be managed. Additionally, orchestration can tie in approval processes at defined points in a given workflow.In an orchestrated environment, the developer would log into a portal (preferably from a device of his or her choosing) and input the desired specifications for the compute environment required. The developer would fill out various fields defining physical/virtual, CPU speed, disk space, performance levels, RAM, etc. The system would then:

  • Verify that any approvals needed were in place, possibly sending alerts for additional approvals to the appropriate parties.

  • Kick off various scripts to provision the storage, network, compute and other resources.

  • Notify all parties upon completion.

    This is a far more streamlined process, but also more complex to create. The last piece that’s truly required to define the architecture as a private cloud is "measured service." Whether or not the feature is heavily utilized, a private cloud architecture should have the ability to show which business units, departments, and consumers are using which resources, and how much is being used. This allows for key business metrics to be drawn, and informed decisions to be made. For example: Department A is utilizing 23% more IT resources than any other department; let’s dig into why. Monitoring and chargeback are key to regulating the value and usage of a private cloud infrastructure.

    With the ability to rapidly provision resources on demand, spin up new applications or expand old ones on the fly, organizations need to ensure the power isn’t being abused. Virtual machine sprawl is an example of this in the server virtualization world. Server virtualization made it so easy to spin up a new server resource that the power tended to be widely overused.

    It’s always important to know what you’re really getting and map that to what you really need. Many organizations will only need some level of converged infrastructure; some will require a full private cloud.

About the Author(s)

Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like

More Insights