Intel: How the data center transforms into a private cloud

As the core part of enterprise IT infrastructure, data center has been confronted with problems in the process of continuous development, such as complicated management, high cost and low utilization of resources. The cloud computing concept, which popped up a few years ago, gives traditional data centers the hope that these problems will be solved if they are flexible and efficient in consolidating their resources and instead operating in the mode of "delivering it as a service."

In previous installments, we talked about energy efficiency, management, and virtualization (click here) from the data center, so the ultimate goal of virtualization is to be a common development direction for the cloud computing hub, which is all data centers. This article will discuss how to turn the data center into a cloud computing hub, and take Intel's own practice as an example to introduce results.

1, four steps to let the data center to private cloud

According to the survey of consultancy firms such as IDC, the current model for CIOs is the private cloud or hybrid cloud. A public cloud like Google or Amazon has many pitfalls for companies: reliability, stability, business risk, etc. So how many steps does a datacenter transform into a private cloud?

At the same time, the data center needs to pool its own infrastructure resources, flexibly allocate resources to related business in an automated management mode, and improve the utilization rate, and can also flexibly deliver computing services to the needs of related workloads. So this year many manufacturers have thrown the data center into the private cloud Four steps: 1, the use of virtualization technology to integrate infrastructure, 2, the implementation of automated management, 3, to achieve flexible workload allocation, 4, the implementation of service-oriented computing delivery methods.

Although there are differences between the two, it can be summed up in the above four steps. For traditional data centers, the most complex is the first phase of virtualization consolidation. Pooling servers, storage, and networks into a resource pool, and so on, and then redeploying the underlying infrastructure of the IT infrastructure, which involves integration systems, business migrations, and more. Many enterprises in China have started to use virtualization technology to integrate their own server resources, in order to improve utilization and reduce costs. But many IT managers equate a private cloud with virtualization as a cognitive myth.

Because virtualization simply consolidates the underlying resources into a pool and does not simplify the management process, the resource allocation of virtual machines is rigid and inflexible. In some cases, even administrators complain about the complex management problems that are caused by virtual machines. Therefore, the second step is to automate the management of the virtual pool resources through the automated resource allocation platform and management system delivered to the foreground user configuration or self configuration. And now many of the domestic virtualization applications also require the administrator to adjust the virtual pool of resource rationing, switch virtual machines and so on. The benefits are not just liberating people, but also dramatically reducing the system's deployment and maintenance cycles.

The flexible distribution of workload can be said to be the advantages of virtualization technology and automation management to a higher level. The system can automatically determine the resources needed for the current business load and draw the corresponding parts from the virtual resource pool-this involves the technology and solution of dynamic regulation, virtual dynamic migration, fine-grained task management, etc. The biggest problem with traditional data centers is the lack of flexibility, whether business expansion requires more computing resources, or the requirement that the release of the corresponding resources is required in the past is a fixed cycle of things to do, thus greatly affecting the production efficiency and cost of the enterprise.

After completing the three steps above, the data center basically completes the conversion for the private cloud. And the fourth step can be said to be the optimization of the work, service-oriented computing delivery methods so that the front desk users and backend managers do not need to carry out cumbersome human intervention or command line. Directly in the form of a Web page or graphical interface to request the data center to provide corresponding computing resources (storage resources), and the data center automatically according to the calculation of the load capacity to divide the corresponding resources. After implementing a service-oriented transformation, the user data center already has the capability of a private cloud, and in the future, a hybrid cloud transformation can be realized.

2. Results of Intel Cloud computing strategy

As we've described in the previous two issue, Intel began consolidating its data center with virtualization technology in 2005, consolidating 150 data centers to 95 by 2009, and plans to consolidate it further by 2014 to 8. After virtualization consolidation, Intel uses cloud computing technology to deploy computing resources flexibly around the world, and to deploy common business in some businesses to private clouds.

Intel's main investment, according to data from the 2009 Intel It performance report, has created a private cloud environment for processor design and testing for the world's major research institutes. In addition, Intel also deploys less risky non-critical business or applications in a private cloud environment, resulting in a further reduction in data center efficiency and it costs.

Intel says it uses virtualization consolidation for data centers and introduces private cloud patterns. 2008 's facility investment in the data center decreased by 65% compared with 2006, while performance increased 2.5 times times. Can be said to further release the original data center resources and performance. Intel expects the cloud computing strategy to save 650 million of billions of dollars in equipment renewal costs by 2014.