Set Goals
Determine what you want to accomplish with your automation initiative. Do you need self-service machine provisioning? Automated responses to changing infrastructure conditions? How about policy-based virtual machine and application life cycle management?
If the intent is to monitor application performance and spin up additional application servers when demand peaks, the plan is going to be different than if self-service provisioning with departmental chargeback is the primary goal. Formulating a concrete set of objectives is essential to the evaluation phase of your automation project.
Then, since product capabilities vary widely, map that goals list to feature requirements. How easy or difficult it will be to make that match depends on the underlying virtualization and management platforms you're dealing with. A full-featured automation product will need to plug into multiple silos to gather the information it needs to make policy-based automation decisions. This means integrating all relevant server, storage, and network virtualization technologies as well as maintaining accurate licensing and consistent configuration information. Depending on how your network is set up, every one of these resources could be a silo, making integration daunting. Survey respondents with a private cloud strategy underscore that point: 58% say integrating existing IT products with cloud is a major issue.
Because integration can be such a challenge, it's important to delve deep into the compatibility matrix of any prospective product before writing a check. If a vendor provides few hooks, it may be impossible to link policy engines without an absurd investment in labor. How absurd? A 2-to-1 or 3-to-1 investment in expert consulting services or internal staff commitment vs. software costs before the benefits of automation are apparent isn't unheard of.
Why? Again, a lack of standards.
That brings us back to goal setting. When it comes time to dig into the details, automation requires that you have a very clear idea of what you want to accomplish so you can create workflows and processes that are repeatable, consistent, trustworthy, and (this is key) reusable. Take incident response: Say an application server has a meltdown that jeopardizes the availability of a key software system. If the application is in a well-constructed app server farm, other servers should continue to meet client demand, but at a higher load with reduced efficiency. If one goal of your automation initiative is a self-healing response to the loss or degradation of an application server, many variables must be considered before an automatic action is taken. We provide examples of goals and their related processes in our full report.
What You Get For The Money
For large enterprises, the cost of these suites can easily reach six figures and climb to $300,000 or more, depending on the level of automation and customization required and the size of the infrastructure. For your licensing investment, you get a task engine that can react to data by triggering workflows and whatever set of common integrations and functionality the vendor bundles; self-service provisioning Web portals, a virtual machine optimization scheduler, or some basic VM power management policies are commonly included.
What you don't get are workflows specific to your network, application portfolio, and business processes. That's why the single most important step is to carefully scope goals and requirements, identify overlap, and conduct a thorough cost-benefit analysis before tackling an automation project.
Still, it's worth getting started. Survey respondents who have private clouds say they've gotten excellent results in terms of reducing operational and capital expenses as well as managing their IT teams' time. Better resource usage overall, life cycle management, and automated provisioning--all very achievable goals--can easily make the effort and expense worthwhile.
(registration required)
