Modular Data Centers are dramatically shortening the data center design cycle from 18 months to as little as four months by using off-the-shelf, standardized subsystems. The goal of MDC providers is to deliver new capacity within a 3-4 month delivery window, more than 6 times faster than traditional designs. In addition to time savings, such scaling addresses many challenges such as unpredictability of cooling systems, dynamic power loads and uncertainty of reliability.
To reduce the risk of unpredictability of cooling systems, many earlier data center designs were oversized and therefore required spending far too much money. Generally speaking, traditional data centers use power and cooling equipment from various manufacturers. In these cases, and with a uniquely sized data center, system performance was predicted by analysis and management software which was customised for an individual project.
With modular power and cooling plant, end-to-end power and cooling systems are pre-engineered and pre-manufactured from standard building blocks. Because of standardization, the capacity of cooling systems becomes highly predictable. Better yet, users maintain the freedom to choose from these standard building blocks to create a solution that uniquely matches their needs. Because of the low cost, predictability and flexibility of these systems, many predict that standard power and cooling modules will eventually become inventory items.
Designing with standardized modular plant is inherently more scalable than traditional methods. This is because system performance is guaranteed by specification and controls are standard. Management software is also generally standard for the modules. Equipment arrives in convenient pre-packaged modules, such as skids, containers, or kits. There is of course some extra installation cost, for example in using 10 smaller extractor pumps instead of one big pump, as well as container and skid package cost, but these costs are more than offset when one considers the costs savings by avoiding custom engineering, equipment and power costs from engineered over-capacity, programming and configuration, shipping and installing damage (5% of data center components are dead on arrival), factory instead of field labour, and more field testing because custom solutions must be integrated and tested on site.
Modules may be internally fault tolerant, and can be combined to achieve redundancy. Furthermore, with modular data centers, there are no extra operating costs but there are operating cost savings from reduced expertise requirements and energy costs, and no maintenance costs on capacity not yet required. These savings are not speculative – they are common sense. The costs of optimized control and management software must be spread across many data centers, and standardized modular plants are ready to handle this challenge to allows this.