In the past few years, data center managers have intensified their focus on optimizing facility infrastructure so that efficiency, performance and reliability are all factored into the health of operations. One of the top challenges is knowing what types of data are useful in guiding facility managers as they attempt to balance what sometimes seem like competing priorities. For instance, it’s a given that operators will need to measure rack kW, since identifying power consumption is crucial for measuring power usage efficiency (PUE). But when the time comes to optimize performance and efficiency on the facility side of data center operations, management will need a more circumspect understanding of their data center environment, and how power load,
CPU and other process shifts will impact efficiency, performance and reliability.
This guest blog post by our Partner Geist looks at three of the key environmental benchmarks that can be used to improve overall facility health.
Last year, Green Grid unveiled a new metric known as Performance Indicator (PI) with the purpose of more effectively measuring the performance of data center cooling systems. The benefit of PI – which is supplementary to PUE – is that it aims to quantify the usefulness of current cooling capacity rather than writing it off as a necessary evil. This makes it possible to estimate how certain changes to data center infrastructure will impact facility health.
That said, to work as intended, temperature metrics would ideally be collected at as many points as possible, and no fewer than three temperature sensors per rack, according to Data Center Knowledge Editor-in-chief Yevgeniy Sverdlik. In general, granular temperature monitoring is a best practice, if for no other reason than that it paints a vivid, real-time picture of facility health. However, with the inclusion of PI in PUE, detailed temperature monitoring can supply the precise and accurate information that is needed to understand more complex, longer-term relationships between the critical variables contributing to facility health.
The key issue concerning humidity in the data center is that there needs to be a bit of a Goldilocks effect. Too much of it can cause condensation, which will ultimately corrode equipment. Conversely, too little moisture in the air will induce electrostatic charges that can damage sensitive electronics. Striking the right balance requires precision measuring of relative humidity at multiple points within a rack.
Unlike absolute humidity – which merely measures the amount of moisture in the air, regardless of temperature – relative humidity will tell data center managers the percentage of the amount of water vapor that can be held in the air at a given temperature. This establishes a measurable relationship between temperature and humidity within a single metric, which is important for facility managers as they raise or lower cooling capacity for efficiency and or performance purposes.
3. Dew point
“Precise planning requires precise information.”
Air temperature may reach a point where, depending on the amount of moisture in the air, water vapor will form into condensation. This temperature is known as the dew point. If the air temperature exceeds the dew point temperature (what relative humidity measures), water will begin to accumulate on data center equipment in small droplets. Enough of this surface liquid can corrode equipment and reduce its lifespan.
In this way, dew point is a moving threshold for relative humidity. The closer it comes to being crossed, the greater the risk of condensation. So, like temperature and relative humidity, dew point should ideally be measured in more than one location on a rack.
Generally speaking, smaller-scale sampling throughout the entire facility is the way to go when it comes to data center monitoring, whether that’s dew point, relative humidity or temperature. Precise planning requires precise information. The purpose of data center monitoring is to collect that environmental information, make sense of it, and then use it to make smarter decisions that will improve overall facility health.
What is your Data Centre PUE? Try our free PUE calculator here.