With 44% of an average data center’s cost of ownership going toward airflow and power usage, you’d think the case for monitoring both the airflow and power in the data center would be obvious, right? Unfortunately, you’d be wrong. It’s staggering the number of facility managers that, when asked about their IT load, have no clue. This is because they are neither monitoring nor metering. The fact is, you can improve your airflow situation until the cows come home. But if you aren’t continuously keeping an eye on both airflow AND power, all the upgrading in the world won’t stop problems from returning.
Get a Clue
I mentioned above that many managers CEG meets at initial airflow assessments can’t tell us what their IT load is. Many question why we’re even asking. The reality is that IT load is a great indicator of both airflow and power efficiency – or inefficiency – and you can’t get it without monitoring.
All power used by a data center turns into heat. The IT, or heat, load represents the number of kilowatts a data center, aisle, or rack is consuming. To illustrate, let’s assume your server rack is using five kilowatts of power. That also means there’s five kilowatts of heat coming out of that rack. You need to provision properly for that heat. To do this, identify both the intake and exhaust air temperatures (air going in and out of the rack), then match cooling as close as possible to the IT load. If you notice a high exhaust temperature (somewhere in the low 90s), you’re under provisioning for that area and need to add more cooling. Likewise, an exhaust temperature in the low 80s reflects overprovisioning. All that being said, if you don’t know your IT load, you can’t correctly provision. And if you aren’t monitoring, you can’t figure out the IT load.
I was at a site last week that was overprovisioning for one row while others were suffering. CEG determined this by monitoring that aisle’s power usage and inconsistent exhaust temperatures coming from the racks. We correctly provisioned for the environment by matching cooling coming out of the perforated floor tiles with the IT load. We confirmed our actions were right by monitoring exhaust air temperatures. All of this was done during an initial data center evaluation using our portable monitoring tools. While this was a good change, we could have assessed the situation even more accurately and recommended more if they had been monitoring before our arrival. The data acquired by measuring a trend over time is 100 times more valuable and useful than just a “snapshot.”
So Now What?
You’ve assessed and remediated your environment. You’re also monitoring and metering. Because of all this, your IT load numbers are at your fingertips. How do you now quantify and qualify success?
Quantification is easy. First, ensure your servers/IT equipment are operating within ASHRAE’s standards. Next…look at the electric bill. Is it lower? Success! It’s just that simple.
Qualification is a bit more complex, but in CEG’s opinion, much more important for gauging success. Qualification is about effectiveness and redundancy. How well is the data center cooled and operating? And in the event of a failure, will the back-up plan enable the data center to run just as effectively? If you can successfully qualify the effectiveness of airflow remediation, the money savings and subsequent ROI will follow.