Data center plans are frequently vague because they can’t account for the constantly changing nature of technical developments in IT equipment. Moreover, it’s close to impossible to accurately predict what lies ahead for a business, so predicting future IT needs presents a significant challenge in the data center planning process.
One option for dealing with some of this uncertainty is the “stepped phase-in,” which refers to an incremental build-out strategy for the system capacity plan during ramp-up of the IT load. The stepped phase-in relies on two simple, intuitive principles:
- Stepped phase-in allows power and cooling capacity to increase with the IT load, eliminating capital and operating costs (particularly the energy expense) of equipment that is not immediately needed.
- If the future IT load is unclear, each step offers an opportunity to re-evaluate the plan and decide whether to go forward, hold off on the next step, adjust the plan down or halt the buildout entirely.
The concepts behind the stepped phase-in are common to everyday life – get what you need as you go, re-evaluate your needs along the way. But the tradition of non-scalable power and cooling architecture has for decades driven upfront buildout of entire data centers. A new approach that incorporates modular system architecture enables designers to take advantage of significant benefits of incremental buildouts, resulting in less wasted capacity during ramp-up, mitigation of growth capacity, and informed evolution of the cooling architecture.
Three factors make the stepped phase-in particularly appealing:
- Energy has become a major expense. Avoiding excessive electricity costs is both desirable and responsible. A data center sized so that the power and cooling capacity track the growth of the IT load is much more efficient than one that has excessive capacity. In addition, “green” building initiatives and “demand-side” management programs provide further incentive to keep costs down.
- Data centers rarely build out to their projected maximum size. Research indicates most data centers never reach the maximum load they were predicted to carry during the planning process. By using a stepped phase-in process, the risk of installing capacity that will never be used is significantly diminished, saving on capital costs – one of the greatest benefits of incremental deployment.
- Unused capacity generates unnecessary maintenance costs. Once it has been installed, equipment must be maintained and repaired even if it’s unused. Considerable service expense can be avoided by installing only what is needed to support the current load. Holding off on installation means not having service expenses for equipment you don’t have.
There will, of course, still be infrastructure elements that cannot be scaled and must therefore be installed for full capacity up front. For example, the physical room size, the electrical service entrance capacity and pre-existing room-based air conditioning are “hard” capacity constraints that should be installed at the outset to accommodate the maximum load anticipated during the data center lifetime.
Scalable elements, however, should be installed to support the initial load then ramped up over time according to the steps of the phase-in plan. Examples of scalable elements include racks, rack-based power protection and distribution, and rack-based cooling equipment.
By using incremental phase-in steps, the user has the option to delay, adjust or cancel full build-out based on actual conditions as they develop during ramp-up time.
For more information on stepped phase-in, see APC by Schneider Electric’s white paper, Data Center Projects: Growth Model.