In any given data center, it’s to be expected that the IT load will change over time. Growing companies typically need more computing power and storage capacity, which of course drives up power consumption. On the other hand, many companies are investing heavily in virtualization technology to reduce the number of physical servers in their environments – and are saving on power as a result.
In either case, the changes affect the power density of the data center and its corresponding cooling requirements. The question is, how best to deal with these changes.
You can’t expect to be able to change power distribution and air distribution equipment in response to every change in the IT load. For one thing, it’s not worth the risk. It’s well documented that human error is the primary cause of downtime in data centers and that changes made to production, operating equipment is often the culprit.
For this reason, it’s a data center best practice that power and cooling distribution equipment for a row or zone be installed and then not changed or reconfigured during the operating lifetime of that row or zone.
In practical terms, living up to that best practice means having a deployment strategy along these lines:
- Lay out rows of racks and enclosures on the floor plan using standard aisle spacings.
- Determine the design density specification for a row and then build a complete row to support that density specification.
- If equipment is to be deployed that is within the parameters of the design specification for an existing unfilled row, it may be deployed in that row.
- If equipment is to be deployed that has a substantially different density than that of an unfilled row, construct a new row designed for the higher density; don’t modify the power or cooling systems in the unfilled row.
- Over time, target lightly populated rows for a complete tear-down and rebuild at a different density specification that is more consistent with current needs.
While this strategy does introduce some constraints in that the power and cooling distribution for a row system does not change after installation, it’s worth the tradeoff in minimizing the opportunity for human error related to changes taking place on live, operating rows.
On the other hand, there are some power and cooling distribution products on the market that permit reconfiguration of the power and cooling architecture without the risk of downtime. For example, the APC by Schneider Electric InfraStruXure system allows for:
- Changing UPS power output by adding hot-pluggable modules
- Changing the type and capacity of receptacles in a rack via hot-swappable rack PDUs
- Adding supplemental cooling air flow capacity to a rack via plug-in rack-mounted devices
This type of equipment allows for some additional flexibility after installation and is particularly beneficial in smaller installations where staged row deployment is not feasible.
To learn more about how to properly plan for data center power capacity, read the
APC by Schneider Electric white paper, “Guidelines for Specification of Data Center Power Density.”