If it is your job to translate overhyped demands to take your business ‘To The Cloud!’ you know there is not enough reality in cloud computing. You cannot start from scratch, nor can you simply deploy dynamic virtualization and call it done. You must accommodate legacy investments, architectural spaghetti, ‘technical debt’, manual processes and more. So where do you start?
In our recent book, Visible Ops - Private Cloud: From Virtualization to Private Cloud in 4 Practical Steps, my co-authors (Kurt Milne, Jeanne Morain) and I spoke with dozens of IT leaders about their experiences building their own ‘private clouds’. By documenting the successes and failures common to the best performers, we came up with a realistic stepwise process that builds on legacy investments, capitalizes on existing skills, and incorporates necessary processes, to deliver the benefits of cloud computing.
Phase 1: Cut through the cloud clutter
The first step entails planning and communicating objectives, managing initial proof of concept efforts, and developing competency roadmaps.
Successful cloud implementations result from executing a business strategy, not rolling out new IT projects. You need to cut through the hype by establishing a service portfolio view of infrastructure and applications, measuring current service performance and cost, setting goals for service improvement, and establishing some initial success, before you start transforming virtual infrastructure into private cloud.
Understanding application performance and response times, service fulfillment cycles, service level metrics, key competencies, operating and capital costs, etc. allows you to plan achievable improvements. This in turn helps to cut through the hype in order to show your business what they should realistically expect from your private cloud strategy.
Phase 2: Design services, not systems
With a plan in place, start to design business optimized cloud services, enable one-touch service ordering, and implement a repeatable approach for build and deploy.
Business services must be standardized, cataloged, and automated to establish repeatable user-driven onramps to deploying resources. This requires a new approach to Business Service Management to avoid an ever-expanding complex catalog of bespoke ‘services’ that are never deployed the same way twice.
This is a critical difference between building virtualized applications and delivering cloud services. IT-centric approaches that elevate administrative complexity and control will not work in dynamic cloud environments. Some essential aspects of legacy BSM frameworks remain important, but cloud computing will kill complex controls in favor of simplified enablement that puts business users in charge.
Phase 3: Orchestrate and optimize resources
With service design complete, you should update monitoring and alerting, codify policy-based event responses, and automate resource changes and workload moves.
Technologies like application performance management, resource optimization, and process automation are immensely important in a private cloud environment. An effective private cloud relies on technologies that monitor real-time performance of end-to-end business services, detect variations from defined performance models, diagnose the true root cause of problems, match performance requirements to available resource pool capacity, and automatically adjust and optimize resource allocation to match.
This is much more than just response time measurement and live migration. Effective private clouds optimize complete business services, not just virtual machines. Live migration is important, but not sufficient, to deliver a successful private cloud.
Phase 4: Align and accelerate business results
With the heavy technology lifting done, complete the transition to a resource rental model by reshaping consumption behavior and streamlining response to business needs.
This entails moving targeted workloads to your private cloud to leverage its benefits, understanding and communicating the service cost, quality, and agility measures of each cloud environment, and actively reshaping demand for IT resources using a rental model.
This change in business behavior enables private cloud to be successful in ways automated virtualization cannot. Virtualization is an IT-centric technology that does not require business users to change their behaviors, as IT is still in charge. With cloud computing, business users are in charge, so they must ‘learn’ some of the discipline needed to maintain acceptable cost, security, risk, performance, etc.
Summary
This is of course a simplified version of the practical four-step process from virtualization to cloud. Clearly developing and delivering your own private cloud is not even this simple. However, with a concise, practical, and realistic approach born of the real-world successes and failures of those who have already done it, as documented in Visible Ops - Private Cloud: From Virtualization to Private Cloud in 4 Practical Steps, you can achieve phenomenal results, drive IT efficiency, and deliver significant business benefits with your own private cloud.
About Andi Mann
Andi Mann is Vice President of Strategic Solutions at CA Technologies. With over 20 years’ experience across four continents, Andi has deep expertise of enterprise software on cloud, mainframe, midrange, server and desktop systems. Andi has worked within IT departments for governments and corporations, from small businesses to global multi-nationals; with several large enterprise software vendors; and as a leading industry analyst advising enterprises, governments, and IT vendors – from startups to the worlds’ largest companies. He has been widely published including in the New York Times, USA Today, CIO, ComputerWorld, InformationWeek, TechTarget, and more. He has presented around the world on virtualization, cloud, automation, and IT management, at events such as Gartner ITxpo, VMworld, CA World, Interop, Cloud Computing Expo, SAPPHIRE, Citrix Synergy, Cloud Slam, and others. Andi is a co-author of the popular handbook, Visible Ops – Private Cloud; he blogs at Andi Mann – Übergeek, and tweets as @AndiMann.
Related Links:
12 Things You Need to Know About Application Performance Management in the Cloud
