Automation is "the use of machines or technologies to optimize productivity in the production of goods and delivery of services. The correct incentive for applying automation is to increase productivity, and/or quality beyond that possible with current human labor levels so as to realize economies of scale, and/or realize predictable quality levels.[1] “

Automation in the private cloud is no different. Its purpose is to optimize productivity in the delivery of Infrastructure-as-a-Service beyond what human labor can provide. It seeks to use the technology in a private cloud (virtualization, monitoring, orchestration, etc.) to provide virtual resources faster than a human can at a predictable level. This, in turn, improves the economies of scale for a private cloud.

Why is automation needed in a private cloud?

Automation is needed in the private cloud because it improves the bottom line (through a reduction of costly human labor) and improves customer satisfaction (through a reduction in the amount of time it takes to complete requests).

Without automation, you do not have self-service and providing self-service is one of the most compelling and cited reasons for a private cloud. Self-service does not refer to entering information in a portal that in turn creates a work ticket or assignment for an administrator to create virtual machines for the requestor. Self-service must be driven by automation to meet the requestors' needs without any further human interaction. The speed by which the request is satisfied is merely limited to the speed of the physical resources, e.g., how fast the disks spin and how fast the bits move across the network. Even on the oldest hard drives and the slowest network, it will still produce the desired result faster than any human interaction in the process.

Without automation, you do not have a cloud. Why? The NIST definition of a cloud lists five essential characteristics of cloud computing:

• on-demand self-service,
• broad network access,
• resource pooling,
• rapid elasticity or expansion,
• and measured service.[2]

Although virtualization optimizes resource pooling and elasticity, and the network configuration provides broad network access, automation is what will enable self-service and measured service.

Besides self-service, automation also opens the door for even more capabilities of the private cloud. Examples of automation in a private cloud:

1. Self-service virtual machine provisioning - creating a new virtual machine based on the criteria entered in the request. Think of deploying a virtual machine without human interaction.
2. Self-healing - defined as taking a corrective action when an incident occurs, without human interaction. Think how loss of a service may cause an automatic recycling (or rebooting) of resources to bring the service back online.
3. Application resiliency and availability - burst an application to a new server or a public cloud based on defined thresholds of application usage and performance. Think how the usage of www.nfl.com increases on a Sunday. Automation can be used to automatically increase the resources available to the app, thus improving performance.
4. Power management - reduce the use of computing resources when they are not needed and bringing them back online when they are. Think of a thermostat – the air conditioner turns on when it gets hot, then turns off when it gets cool, all without someone physically turning the power to the AC unit on or off.
5. Chargeback or showback - collect data and report on the usage of the physical infrastructure to help with cost allocation or awareness. Think of a monthly credit card statement to show how much you are spending every month on datacenter resources.

Automating the private cloud has big potential for OpEx savings and CapEx optimization. At the very least, automation should be used to enable or improve redundancy and recovery of private cloud resources to ensure the lights stay on and the applications stay running. It can reduce the outage time of applications (OpEx) and make more efficient use of datacenter resources (CapEx).

OK, I understand the value of automation. How do I enable it?

At a basic level, automation can be enabled with runbooks. Runbooks can take a series of pre-assembled actions (like scripts) and execute them when a trigger occurs. The automation is able to detect the trigger, activate the runbooks, run the scripts in the defined order, and then report the result of the runbooks.

A basic example:

1. Trigger - server goes offline and cannot be pinged.
2. Runbook – detects the trigger has been fired and activates a series of scripts, in a specific order, to perform.
3. Scripts (examples):
   a. remove server from the load balancer so it cannot respond to any application requests
   b. e-mail the application administrator of the outage
   c. attempt a forced reboot the server to attempt to bring it back online
4. Results - did the scripts succeed? did a forced reboot of the server bring it back online? or did it still fail? If server is online, e-mail admin again. If server is offline, open an incident in the IT service management system.

In essence, the automation eliminated manual (human) triage steps and quickly brought the issue to an admin for either notification that the problem was automatically fixed or raised an incident for an admin or engineer to begin troubleshooting.

At a more advanced level, you can enable policy-based automation that can perform certain tasks based on the identity of the user (requestor). For example, when requesting virtual machines, the user is limited to only 5 VMs, because of membership in the testing group.

How do I ensure my automation efforts are successful?

There are several recommendations when automating a private cloud:

1. Determine the goal of automation and what's important to you. Is it to reduce human action in certain situations? Is it to increase availability and performance of applications? Is it to enable faster outage triage? Is it to meet service level agreements? what do you want to achieve with automation?
2. Include automation from the beginning. Do not build a private cloud, and then try to retrofit it with automation at a later time.
3. Automation efforts must include computing, storage and network resources for it to be effective. Including 1 or 2 out of the 3 will not yield the more optimum results.
4. To increase the value of automation in the private cloud, it should interface with existing business systems, such as license management, IT Service Desk, escalation procedures, etc. to provide a more holistic lifecycle around the automation touch points.

Summary

No product provides the runbooks necessary to automate your environment. This is the investment needed to realize the benefits of automation - making it real in your environment. Use the automation toolkits to define business or IT processes and specifically, the triggers, scripts and results reporting.

One product that delivers the tools to enable private cloud automation is Microsoft System Center. Automation is built into each component of System Center, with Orchestrator acting as the runbook automation engine. Each System Center component is instrumented with a PowerShell interface and an Orchestrator Integration Pack. These go a long way in helping you develop and test your automation. There are also Integration Packs for popular components from HP, IBM, BMC, VMware, EMC, NetApp and much more. This will ensure that the automation can reach out across the IT infrastructure for a more comprehensive solution.

For more information on System Center automation capabilities, click here.

[1] http://en.wikipedia.org/wiki/Automation

[2] http://www.nist.gov/itl/csd/cloud-102511.cfm

Thanks!
Michael Lubanski
Americas Private Cloud Center of Excellence Lead
Microsoft Services
mlubansk@microsoft.com