It’s a debate that’s raged on for years: Which is cheaper, public or private clouds?
A new report from 451 Research finds that two of the most critical factors that influence the cost of a public versus a private cloud deployment are an organization’s ability to efficiently manage infrastructure and utilization of hardware resources. Generally speaking, if any organization has the expertise to manage a large number of servers at a high level of utilization then on-premises, customer-managed private clouds can have a total cost of ownership (TCO) advantage compared to public clouds. For smaller environments, or any sort of variable workload demand, public cloud is a more attractive financial option, 451 Research’s Director of Digital Economics Owen Rogers reports in “The Cloud Price Index: The great public vs private cloud debate.”
Public and private IaaS clouds each have their pros and cons. Private clouds come with an inherent control of the infrastructure, which gives some security-conscious and regulatory-sensitive customers peace of mind. On the other hand, true private clouds that allow for virtualized resources to be self-provisioned by users can be difficult to set up and manage. Public clouds, meanwhile, provide access to an almost infinite amount of infrastructure resources without any upfront investment required, and the ability to use cutting-edge technology available first from public cloud vendors. The perceived lack of control and security can be a big turn off for some customers though.
Financial comparisons of a public vs. private cloud are another way to consider the issue. To determine TCO, Rogers recommends that an organization consider how large of a cloud deployment they will have, how efficiently they can manage it, and how much it will be utilized. A key measurement in determining efficiency of a cloud is the number of virtual machines managed per engineer. Generally speaking, if an organization is able to achieve about 400 virtual machines under management per engineer, then a private cloud can be a more attractive financial option compared to a public cloud. If the organization is not able to reach that efficiency level, then public cloud can be more efficient.
Another key consideration is utilization of those resources. If the infrastructure is only used to about 50% of its capacity, then it requires a cloud administrator to manage up to 1,000 VMs each to achieve a TCO advantage compared to public cloud. “It is possible for self-managed private clouds to be cheaper than public cloud, but utilization and labor efficiency must be relatively high,” Rogers finds.
If organizations do pursue a private cloud option, Rogers also analyzed an open source OpenStack option compared to proprietary offerings from VMware and Microsoft. Generally speaking, if an organization has high utilization of resources and less than 400 VMs per engineer, then VMware and Microsoft have a TCO advantage. If an organization has expertise to manage more than 400 VMs per engineer, then OpenStack can be a more attractive option. Rogers recommends in most cases that using a commercial distribution of OpenStack is more financially advantageous compared to a do-it-yourself approach of using the open source.
Another metric to consider in the OpenStack vs. proprietary private cloud decision is the cost of labor. Rogers estimates that an OpenStack engineer average salary is $140,000 compared to that of a VMware or Microsoft administrator’s $100,000. New training and certification programs from OpenStack should increase the talent pool of available open source cloud engineers, which could even out the salary discrepancies, Rogers also notes.
Public clouds have their advantages. Most notably, public clouds are the least wasteful deployment option because they offer on-demand provisioning of resources. This frees the customer from investments or capacity planning exercises that are critical to a private cloud deployment. Cloud vendors do offer customers discounts for committing to longer-term contracts of use, however.
Rogers warns that a private cloud deployment can turn into a financial quagmire if utilization and efficiency rates are not achieved. It’s difficult to predict how many engineers it will take to manage a fleet of VMs and how well utilized it will be within an organization. “These (unknowns) are real risks – get them wrong, and a black hole effect can make resources disproportionally expensive,” Rogers notes. There are other X factors to consider as well, including cost of migration, existing infrastructure investments, in-house expertise of systems management and potential training.
TCO is also just one factor, he notes. “Often the security and control inherent in private clouds outweigh financial considerations,” Rogers reports, adding that a long-term strategy should be considered to determine what the future path of the company is and which platform offers the key features most important to them.