Data center performance is much more than its PUE and must account for asset utilization, total energy costs and carbon usage effectiveness (CUE). I am sure most CIO’s and data center managers already know this from their efforts to reduce costs while also meeting goals for IT sustainability, IT asset utilization, and virtualization.
Managing data center energy consumption is definitely important and a great way to produce budget savings. And, PUE provides the standard method for measuring data center energy performance. But I wanted to revisit my last post on PUE Version 2 because I thought a little more information might be useful especially for those having a broader sustainability agenda.
Improving PUE Results
PUE = Total Data Center Facility Energy / Total IT Energy
To improve PUE IT departments should focus efforts on both aspects of the metric – reducing the non-IT facility load and reducing the total IT loads in the data center.
To reduce facility loads look to:
- Raising the operating temperature in the data center following ASHRAE TC 9.9 expanded operating bands
- Change to more energy efficient fans in racks and adjust fan speeds
- Install occupancy sensors to control data center lighting or consider low voltage alternative lighting technology
- Follow best practice for sealing air leaks and maximizing proper air flow
- Conduct energy audits and remediate inefficiencies
To reduce IT loads look to:
- Decommission and eliminate ghost systems that are no longer needed
- Maximize virtualization to eliminate physical systems not just servers
- Use capacity management to eliminate underutilized systems
- Implement converged infrastructures as part of a virtualization strategy
The PUE Paradox
You don’t have to be a math wiz to see that some efforts to reduce total energy consumption will have a negative effect on your PUE. Likewise you might have a better than average PUE but still have lots of idle systems running.
You can see some of this dynamically using the PUE Calculator created by 42U in conjunction with their Energy Efficiency Calculator. Using these two tools, CIO’s can get immediate estimates for the benefits of various projects. That includes being able to compare one data center location to another based on typical state energy costs.
That’s why it is important to realize that simply focusing on your PUE without regard for total energy consumption, cost per kWh or system utilization can be very misguided.
Similarly, that’s also why you have to be careful not to engage in PUE wars that compare one data center with another data center purely on the basis of their PUE. That’s because there are several other factors that can effect PUE such as climate and levels of reliability.
CUE Versus PUE
Carbon usage effectiveness (CUE) was developed to prepare data center managers for the emerging importance of carbon emissions (re: consumption) and the growing interest in carbon taxation. CUE gives data center managers an IT sustainability metric that enhances PUE when they are used together by accounting for the carbon emissions ’caused’ by the data center energy consumption.
There are two methods for calculating CUE. The first method uses the same value for Total IT Energy used to calculate PUE but uses the Total CO2 caused by the Total Data Center Energy.
CUE = Total CO2 Total Data Center Energy / Total IT Energy
And the alternate method of calculating the CUE is to multiply the Carbon Emission Factor (CEF) by the PUE of the data center. You can find the CEF for your region using the US Energy Information Administrations data.
CUE = CEF x PUE
When combined with PUE, CUE gives data center managers a more well rounded view of data center performance from a sustainability perspective. More importantly, CUE and PUE give data center managers the ability to consider alternative energy sources to power the data center which can be extended to analyze alternative sites for locating a data center or prioritizing green data center initiatives.
Pulling it all Together
The priorities for your data center performance strategy might be hard to define with so many competing demands. There’s the goal for reducing total energy demand which drives total cost as well as carbon emissions making it a prime candidate for the top priority.
Then there is IT asset utilization and the need to maximize returns on capital investments which could be a close second. If you still have a lot to do on virtualization and converged infrastructure your IT asset utilization might be your top priority.
Finally, you may also want to leave some room for increasing the use of alternative energy sources for your data center energy consumption. So if you have a choice of data centers for consolidation or colocation providers the availability of alternative energy sources may become another factor in your sustainability priorities.