Well, perfect needs to be defined. But in general, there are major factors involved that serve as the building blocks of a reliable, resilient and robust data center:
1st and foremost is "comprehension". You cannot build a data center without understanding the business needs. Now a lot people talk about this but don't really know the importance nor the scale of it. What's the purpose of the data center, are you building it for in-house use or co-location/managed hosting? If in-house, are you acting as a self-sustained department and a profit-center? In other words, do you invoice your departments for hosting? What is the cost of your downtime? What's your affordability time (cost) wise for failure? Just to give you an example and to elaborate how important this factor really is, a "perfect" data center for an organization might be a Tier I DC with in-rack cooling and a DR sitting in a container at a remote location while a "perfect" data center for another organization might be a Tier IV DC with chilled CCUs, cold-aisle containment and a buildup out of a green field! A perfect data center in our view is what suits organizational/business needs perfectly, by all means be it techno-commercial, real estate, operations, or what have you. In short, it must be perfect for the person who is paying for it, not the UPS salesman.
Next is proper planning of capacity, size, scope, project, budget, resources, assets, etc... A lot goes back to admitting the fact that we really can't do all this in-house! We need proper consultants who could deliver proper engineering, design, bill of quantities, project QA and supervision, training, etc...
Based on that, the "design" is the basis of the data center. Organizations typically think this is a one-page layout. Design is a full "kit" of detailed designs and specifications driven out of a conceptual framework. It needs to be validated and verified by the applicable standards and best practices, then put into prospective with respect to budget, time, delivery and other implications, and once finalized then detailed and finally consolidated.
Technology selection is another factor. Today you can build a Tier IV data center that is a decade old, while you really need to build a data center that is a decade ahead of its time because you are building it to last you for at least a decade! So in our view, today's technology is not good enough. Thinking about how many times the IT equipment such as servers will be renewed in the life-time of a data center should make people think twice about the infrastructure they are putting in place. Is it reliable enough? Is it robust enough? Is it scalable? If yes, is the scalability cheap enough? Operational and maintenance issues all must be evaluated and factored in.
Supervision plays a major role in implementation. Testing and Commissioning is another major one. Many people skip the "core" objective of this task. Be it for lack of subject matter expertise or being hurried to go live as soon as possible they skip the fact that testing and commission should validate and stress-test the capacity and resilience of the data center as a lump sum integrated piece against the original design and specifications. Many customers just move in their IT, go live, inevitably grow in the facility and by the time they reach the stress point, they find out that the actual build up is not but 40% of what it was intended (designed) for. That’s years down the line where warranties are not applicable anymore, and consultants and contractors are not available and any minor move or adjustment to the infrastructure would just be too much.
Localization is another factor. That’s why we work through channel partners and posses a wide reseller network. You need to be local, period! Our channel partners are carefully handpicked and selected. They come from diverse backgrounds with vertical-wide and supply-chain management expertise are fully supported and trained by us. They are supplied with full hands-on and certification training, complete assistance on the job and during a project’s full lifecycle, and we don’t just let them be. They are constantly monitored on the quality of their staff, tested on their expertise level, checked their quality and delivery through constant customer surveys, etc.
Documentation: having the right operations manuals, DR plans, commissioning and decommissioning procedures, SLAs, OLAs, vendor supports, spare parts, well-trained and regularly drill exercised human resources, proper monitoring, routine tests… all of that needs to be in place. In general there is a whole lot that needs to be put together and all the pieces need to fit perfectly to accommodate true “perfection” in the data center.