You’ve no doubt heard that hot spots are a bad thing in a data center because they cause IT equipment to run hotter than it should, a potential setup for failure. But do you know what the definition of a hot spot is, and the best steps to take to remedy them?
A new Schneider Electric white paper, “How to Fix Hot Spots in the Data Center,” answers these questions and more, providing sound advice on how to deal with a common data center problem. It was written by Paul Lin, a Senior Research Analyst at Schneider Electric’s Data Center Science Center.
First, the definition, which I’ll lift directly from the paper:
A hot spot is not any random hot temperature inside of a data center. We define a hot spot as a location at the intake of IT equipment where the measured temperature is greater than the expected value as recommended by ASHRAE TC 9.92. Hot spots occur most often towards the top of a rack. The recommended and allowable temperature ranges at the location of the server inlet can be found in ASHRAE’s thermal guidelines.
Typically, a lack of overall cooling capacity is not the culprit behind hot spots. Rather, it’s a matter of effectively circulating that cool air. Specifically, hot spots occur for two main reasons: bypass, when cool air can’t get where it needs to go, and recirculation of hot air exhaust to the inlet side of a rack or enclosure.
The paper outlines three main ways to find hotspots:
- Feel it by walking around
- Manual measurements
- Automatic monitoring
While they all work, probably you guessed that the third one is the best method because you’ll be dealing with live data. When coupled with data center infrastructure management (DCIM) software, operators can get actual inlet and outlet temperatures at the device level and alerts when thresholds are reached.
The paper then goes through a variety of options for dealing with hot spots, some of which would be downright laughable if not for the fact that some folks actually use them. Apparently some data center operators have been known to “blow air across ice into the cold aisle,” as the paper says, noting the practice is ineffective. Others place pedestal fans in front of problem racks, which is a bit more practical but, as the paper says, for “temporary emergency use only.”
In all, the paper goes through eight potential solutions, only one of which is labeled as a “good practice.” Most common practices fail to address those two main causes of hot spots: bypass and recirculation.
The paper walks through five best practices for addressing hot spots, namely:
- Manage airflow in the rack – such as by using blanking panels to prevent hot air exhaust through unused U positions.
- Manage airflow in the room, such as sealing all openings in the raised floor, using brush strips to seal cable cutouts behind racks and under PDUs, and using containment systems to prevent the mixing of hot and cold air over the tops of racks and around the end of racks.
- Relocate problem loads by removing IT equipment from any enclosures whose load exceeds the design average value and dividing up the load among multiple other rack enclosures.
- Change the location of the air temperature sensor to the supply air stream rather than the CRAC return air stream, where airflow may be unpredictable.
- Allow DCIM to control airflow of cooling units. Take this step only after implementing effective air management practices.
Learn more about what may be causing hot spots in your data center and strategies for effectively dealing with them. Download the Schneider Electric white paper no. 199, “How to Fix Hot Spots in the Data Center.”