Skip to main content

Building a Dynamic Infrastructure with IBM Power Systems: A Closer Look at Private Cloud TCO

Many companies are finding their need for greater business agility being frustrated by an increasingly costly and rigid IT infrastructure. The culprits are many. Maintenance of the current environment accounts for over 70% of the IT budget, leaving less than 30% available for new projects. Annual operational costs (power, cooling, and labor) of distributed systems and networking exceed their acquisition cost by 2-3X and continue to climb. Utilization rates of these commodity servers hover around 5-15% on average, leading to excess capacity going to waste. Time to provision new servers can be as long as six months, hampering lines-of-business efforts to quickly respond to competitive threats or new opportunities. As a result, LOB units are beginning to go outside the datacenter to public cloud providers like Amazon in hopes of lowering their costs and improving their responsiveness. To avoid disintermediation, IT needs to re-invent the datacenter by moving towards a more dynamic infrastructure. One that takes out cost through the use of virtualization to improve utilization levels with a commensurate reduction in power consumption. One that embraces a private cloud model that uses standardized workloads and service automation to dynamically provision IT services in minutes/hours rather than months (and at lower cost) via self-service portals. Customers can build such an environment using IBM’s Power Systems servers coupled with Tivoli service management software.

This paper examines the Total Cost of Ownership (TCO) for a dynamic infrastructure built around private cloud services and compares it to public cloud alternatives as well as conventional one-application-per-distributed server models. The results show that private cloud implementations built around new POWER7 based servers can be up to 90% less expensive than public cloud options over a three year period and over 70% less than a distributed stand-alone server approach.

Content navigation

Related links