The Five Steps of Building a Data Centre
Data centre construction is booming. Despite the rising cost of materials and real estate, the data centre construction market is expected to grow by $8.7bn over the next four years, achieving a CAGR of more than 10%. More than 100 hyperscale data centres were built last year, with hundreds of more new projects already in the pipeline.
Phase One: Assessment
Once a developer finalises the project’s goals and target customer, the assessment phase usually begins with site selection. Choosing the right location for a data centre can have a massive impact on things like the facility’s access to renewable energy or ability to do free cooling.
Phase Two: Planning
This phase (along with the design process) can sometimes take place alongside or even before the site selection and assessment process. The developer outlines the different parameters the site will need to fulfil, from a target PUE, whether to build greenfield or retrofit an existing site, levels of power redundancy, and whether the site will be built in multiple phases or all at once.
Phase Three: Design
This is where the broad criteria established in the assessment and planning phases solidify into answers to specific questions like how many racks the data centre will have, how the building’s power can meet certain standards like LEED certification, and how individual server halls will be laid out.
Phase Four: Construction & Commissioning
This phase is all about execution of the site’s design, and typically involves extensive collaboration with partners and contractors in order to execute on completed designs. The timeline for building a data centre can vary wildly depending on regulatory, logistical, and budgetary concerns. A simple 2 MW colocation facility could be online in just a few months, whereas a 200 MW hyperscale campus might be under construction for years - which is why many of these campuses come online in multiple stages.
Phase Five: Operation
In order to maximise revenues and begin to create ROI, operators will try to keep the time between the completion of a data hall and the move in date for its customers as short as possible. Once a facility spins up, it also begins generating data about itself, allowing for on-the-fly tweaks and redesigns which can further improve the performance of later stages of the facility.