Using the best practices to control air flow minimizes the amount of wasteful air mixing. Less cooling and air circulation is needed, saving energy overall. The main airflow issues to solve for are: bypass of supply air (air meant to cool the server); and recirculation of return air (hotter air not meant to return back to the server). In an ideal situation all of the supply air would enter the server only and all of the return air would go directly to a cooling device. This works well when sizing cooling equipment and fans on a spreadsheet but reality is much more complex, even when using containment, blanking panels, and other air-separating measures.
Supply air pressure can be too high, causing excess air to be forced through cracks, leaks, or pushed through the servers. And with higher pressure, there tends to be a higher amount of air volume changes and fan energy used. With most of the equipment on variable frequency drives, pushing air that excessive is a waste of energy.
With a supply air pressure that is too low, recirculation of hot air can happen in localized areas or across the entire data center. Whether the air is served from above, below, or across a Venturi effect can occur along a row of racks that pulls air from the hot server exhaust to mix with the supply air, increasing the temperature, reducing the effectiveness, and overall wasting energy.
To balance the over or under pressurization tends to be an entirely separate issue from cooling. Adding accurate pressure sensors in the spaces served can quickly give feedback to respond to current conditions more accurately, allowing the servers to pull the supply air as needed and give a higher air temperature difference between the supply and return, thereby increasing the mechanical cooling system performance.