The mind set of Cloud Computing is intended for public clouds such as Amazon’s EC2 or Smart Cloud Enterprise (SCE) from IBM. But there is also the option of using the resources of virtualization and automation within our servers to create a private cloud. This option is highly regarded in companies that already have a large computational capability, has well-defined security policies and regulations are limited by their business sectors.
A private cloud covers the basic formula of cloud computing, virtualization and standardization + automation + self-service, with the only restriction that covers only the servers belonging to the company. Private cloud solutions generates nearly the same value as a public cloud, except for capital spending, which is exchanged in the public cloud by operating expenses (opex). In a private cloud, company has to invest in the acquisition of their computing resources. However, it can significantly improve the operational costs, because the processes of virtualization, automation, self-service standardization drastically reduce support costs and operation.
In fact, in designing a strategy for private cloud adoption, the company must set parameters on benchmarking data centers of companies in its sector, but the data centers of companies are typically derived from the cloud world. The results can be amazing as it is reducing time-in-deployment of their new systems from weeks to hours or minutes. A development environment and testing of a large corporation, which generally takes weeks to meet a demand can do the provisioning of computing resources for developers in a few hours. Imagine the gain in terms of exploiting opportunity to launch new businesses. Another benefit is the best use of the resource pool. Instead of an average utilization of 10% to 15% on each server can use this quadruple, generating more computational power usable within the same capacity.
But, by doing an autopsy of a private cloud that we see at first glance, it is not much different in a virtualized environment. But virtualization is the only prerequisite. It adds a layer of intelligence (software) that allows us to provision, allocate, monitor and manage computational resources automatically, allowing users to order these resources themselves via self-service portal. The automation of provisioning and management of computational resources enables to greatly reduce the time spent by system administrators in mundane activities (upgrade version of the operating system, for example) and each of them can manage a number of servers at least two orders of magnitude higher than the current number. That is, instead of an administrator for every 100 machines can have one for every 10,000 machines. This layer of automation is creating a private cloud environment.
What is this layer? To implement a private cloud environment, it is necessary to implement technologies that manage access resources (the cloud) by users in the form of self-service (we speak here in identity and access management), automated management of resources (service catalog management, service level management, configuration, etc.), ability to manage chargeback and monitor the performance of the machines and so on. Of course all this technology should be operated under security policies and access control.
And how to create this layer? We do all the work manually, using Open Source technologies. The downside to this approach is the need of the company to develop itself the whole work of creation and integration of cloud environment. Another alternative is to acquire the services and technology of a company that provides comprehensive solutions as like eNlight Cloud provided by ESDS.
OK, private cloud solution is a good alternative. But it applies to all companies? Not always … Companies with small data centers often fail to gain sufficient scale to generate a cost-benefit appropriate to build and operate private clouds. The solution for them, in my opinion, is to migrate to public clouds. Or it may be a start-up creative economy. I believe that companies in this sector should go straight to public clouds. There are also cases where performance issues for the servers cannot be shared and cloud sharing is par excellence. We will also cite cases where applications use very different technologies which contradicts one of the basic requirements of cloud computing is that the standardization of environments. The more standardizable, the greater the economies of scale, because we can select one of a set of images (virtual environments) and operate the application on it.
Finally, it is worth remembering that the cloud environment is flexible and does not need to be all private cloud or public cloud everything. We may use hybrid clouds, serving part of the computational demands of the company in private clouds and public clouds part. In short, cloud is already among us. If we begin by public or private cloud depends on the company’s strategy and its characteristics. But we have arrived at the point where it no longer makes sense to question whether or not to go for cloud, but when and how fast we need to go.