It seems that for everything in life today, we hope that someone has developed an algorithm that will calculate something for us — a magic formula that takes the guess work out of what’s needed.
This mindset came to my notice through an online data centre forum. A contributor asked for a formula that could allocate space to satisfy the various data centre functional areas.
You know, plug in maybe some rack units to represent processing, storage and networking; a power usage figure of n kilowatts; a cooling metric; a tier rating; the need for a certain fire suppression system and areas like a loading dock, staging rooms, etc.; and perhaps a head count of people. Then, abracadabra, out pops a number to represent the space required and how it’s broken up.
Unfortunately, data centres are technical beasts that are engineered according to your business requirements.
They have two primary metrics, namely space and power. Both are proportional to the IT infrastructure that’s to be supported and the facility infrastructure that needs to support the space.
You might start with a quantified area of land to work with and then decide how it may be carved up to satisfy a data centre’s functional areas. These include the data halls (also called white space); the elements of the power train such as transformers, UPS, battery or DRUPS; mechanical elements like chillers, air coolers, pumps, CRACs infrastructure; as well as administrative areas, fire system, security room, foyer and outdoor amenities.
The relationships of such spaces aren’t defined in any algorithm I know of.
It reminds me of the equation that attempts to determine the probability of alien life in the universe, the Drake equation. I won’t go into the specifics of it, but there are many indeterminate variables that can all be changed to arrive at different results. Without constraining the variables with known quantities, you’re shooting in the dark.
An algorithm that will account for every variable in determining a space envelope in a data centre is like the Drake equation: indeterminate.
Even the simple issue of determining the data hall size within a prescribed square metre envelope is open to a whole lot of variables:
- Is the IT infrastructure to be housed in the data hall high-density computing? Or a mixture of low, medium and high?
- What are the form factors for the equipment to be racked (rack unit size)?
- Will the data hall be an open rack plan or will it be contained?
- What other infrastructure may be housed in the data hall?
You might be radically planning an all liquid cooling solution which changes the room’s space requirements. The list of exceptions goes on even before you start on the other functional areas.
The engineered solutions in the data centre market are wide and varied and they could all satisfy one’s needs, albeit in different space envelopes.
But there’s only one way to allocate data centre space: by investigating the business requirements the data centre has to satisfy and translating them into the engineering principles that combine to make up the three dimensional design of the data centre.
