The primary benefit of private cloud computing is speed. Companies would require speed to execute a service through a self-service interface while automating its delivery. Private cloud computing is simply a combination of virtualization and cloud computing.
Companies would do well to deploy private cloud for computing resources carefully since it calls for considerable investments in terms of technology and processes.
Investments in private cloud computing should begin only after the organization has assessed its business goals.
The company's IT team will need to acquire or outsource skills such as managing VMs and VS, management of the private cloud environment and security.
Arun Gupta, Group CTO at Shoppers Stop
There are a few strategic conditions that companies should examine before building a private cloud:
i. Application density: It is advisable to invest in private cloud computing only if your company has sufficiently high number of applications. Arun Gupta, Group CTO, Shoppers Stop, says, “If one physical server is running 10 virtual machines with numerous applications, peak load time will see a contention for resources (CPU and memory) and slowing down of all applications. It is best to distribute the applications across different physical servers and balance with applications that do not require the resources at that point in time.”
ii. Level of virtualization: The level of virtualization is another aspect that needs examining. Companies need to understand the amount of resource utilization existing in their data centers. As private cloud computing deployments leverage virtualization technologies, organizations moving towards this will, therefore, need to have highly virtualized data centers. “They would want to use efficiency and consolidation derived from virtualization to improve business productivity and capability,” says Sid Deshpande, Sr. Research Analyst at Gartner.
iii. Knowledge-base: Investing in private cloud computing implies that the organization has built an in-house resource pool of knowledge about private cloud computing. “The company’s IT team will need to acquire or outsource skills such as managing VMs and VS (virtual storage), management of the private cloud environment, and security,” adds Gupta.
In addition to these, there also are some finer aspects that a company must look into.
Assess if your company’s IT infrastructure has a cloud affinity by factoring in the following aspects.
i. Application testing: Test the applications proposed to be deployed into the private cloud for manageability, scalability, and reliability. Citing an instance, Gupta says, “At Shoppers Stop, prior to adopting private cloud computing, some off-the-shelf applications from ERP vendors worked well during the testing phase. Once in production, scalability was found to be a problem even after allocation of adequate CPU and memory resources. The problem was with the application.”
Gupta offers another example of a packaged solution that failed to work in a private cloud environment. “While those were the early days of virtual servers and virtualization, some applications do not work well on any virtual server, sometimes from the same vendor, even today,” he says.
ii. Interoperability testing: When virtualized applications inter-operate with others on similar platforms, the dynamic scaling up or down of resources may cause malfunctions. This is unlikely with certified applications; but non-certified applications require significant testing. “So do consider interoperability testing while preparing for private cloud computing,” recommends Gupta.
iii. Timing of the rollout: If companies do not face any of the issues given above, it’s time for rolling out private cloud computing. Make it part of a technology refresh program. Adoption of virtual machine offers better results when upgrading or replacing a server as one can justify TCO.
iv. Private cloud SLAs: Service level agreements (SLAs) in case of private cloud computing is controlled by the organization. “As a result, SLAs are negotiable,” says Deshpande. Typically, private clouds are offered as added service models from hosted services providers. User-requested SLAs are defined at this level and used to architect the delivered service, as well as define the levels of proactive management processes such as availability management, performance management, etc.
Gupta concludes with a caveat here. “There is nothing called ‘right deal’ when the technology is as yet evolving,” hinting that differences are bound to occur between the SLA and real business experience.