Intel IT Cloud Computing Best Practices
In a recent whitepaper, Intel IT outlined a strategy to accelerate our virtualization efforts. Last week I had the opportunity to talk with Das Kamhout, Cloud Architect with Intel IT, to understand how it is going. What I learned was equally interesting and impressive. In the past 6 months, Intel IT has doubled the rate of virtualization inside our Office / Enterprise (O/E) computing environment. We have gone from ~12% virtual (Dec 2009) to over 30% today (July 2010). The Intel IT O/E computing environment supports our general purpose computing applications in addition to our e-business, supply chain and other enterprise applications required to run our business at Intel.
During our discussion Das outlined several IT strategies that we are implementing to achieve this accelerated virtualization strategy
- Proactively Partner with Application Organizations. This involves identifying and engaging business owners who have applications currently deployed on physical hardware. We work actively with them to drive alignment and confidence to move their applications to a virtual environment.
- Drive New Landings on Virtualized Servers. As new service requests are made from business owners, we now proactively landing these applications in virtualized environments. Our midset has shifted dramatically. Previously we challenged ourselves to prove why we should virtualize. Now we ask the question, "Why shouldn't we virtualize this application.
- ID Barriers & Implement Technical Solutions. As we drive to achieve strategies #1 and #2 above, we do find that there are technical barriers for some applications that can limit our capability or willingness to virtualize. So our 3rd strategic emphasis is to identify these situations and put IT solutions in place. (look for another blog shortly on this subject)
- Execute Efficient P2V Migration. Once we identify the P2V (physical to virtual) migration opportunities (strategy #1) without technical limitations (strategy #3), we want to drive the migration quickly and painlessly. Here we are applying a factory mentality that moves the application without issues and effectively EOLs the older server in the process.
- Proactively Plan and Manage Capacity. The goal here is to put IT metrics and tools in place that can project and compare the demand for Virtual Machines against the available physical server capacity in the required locations. Since we can land new VMs much quicker than we can purchase new hardware, we must manage compute capacity more closely from both an allocation and consumption perspective
As an ex-supply chain manager at a previous company, the most intriguing aspect of our virtualization strategy and implementation of an enterprise private cloud is in our terminology and metrics. The IT and business terminology we are starting to use internally to manage our compute, storage, networking infrastructure is remarkably similar to the terminology that companies use to manage efficient and effective supply chains … weeks of inventory, time to market, forecasted demand, service level agreements and agility.
I invite you to read the Intel IT whitepaper titled “Enterprise Private Cloud Architecture and Implementation Roadmap” for more information on our cloud computing strategy.
Look for more of Intel IT’s experiences, best practices and lessons from our virtualization implementation in future posts in our IT community. I'm expecting Das to weigh in with his insights and expertise soon in this IT community.