Reducing Energy Consumption and Cutting Costs in Shared Data Centers

By Lars Strong, P.E., Senior Engineer, Upsite Technologies

Many types of companies use colocation facilities - shared data centers - because they don’t have enough cabinets to justify their own data center, or they want the higher reliability of a colo site for their most critical equipment, even if they already have a data center of their own. Colocation facilities are no different than any other company that requires reliable power, connectivity, and cooling to run their servers. One of the largest expenses of a facility, yet arguably one of the most important for equipment protection, is cooling.

Cooling is especially important to the owners of a colocation facility because it is a large percentage of the total operating expense. The costs associated with leasing space in a colo facility are important to its tenants, and also important is the knowledge that their servers will be protected and kept running. Without proper cooling providing consistent and appropriate intake temperatures, the reliability of their servers will be compromised.

ASHRAE (the American Society of Heating, Refrigerating, and Air-Conditioning Engineers) sets standards for IT equipment intake temperatures. The recommended range is 64°F to 80.6°F (18°C to 27°C). These parameters are set mainly to achieve the most reliable operation of IT equipment and secondarily to allow for the efficient operation of the cooling infrastructure. Keeping an ever-changing facility environment within these boundaries, however, can be difficult to maintain.

The challenges colos face when trying to maintain best practices for airflow and thermal management are primarily because of having multiple customers in the same room. Cages required to provide security make it hard to maintain hot and cold aisle layout and prevent spaces in cabinet rows. It can also be hard to make customers use blanking panels, and hard to control placement of IT equipment in cabinets to assure front-to-back airflow. All of these issues are related and together can significantly compromise air flow management (AFM) best practices, the prevention of mixing and recirculation of air.

The best way for colos to address these issues is in the initial customer contract/agreement/SLA (service level agreements). This is where the colo can require customers to use blanking panels and adhere to facility requirements, etc. Encouraging and enforcing AFM best practices are much harder to do if they’re not included in the original agreements. For colos that want to do as much as they can to reduce OPEX and improve profitability, here are some steps that can be taken:

Calculate the CCF

The Cooling Capacity Factor (CCF) was developed by Upsite Technologies to help data centers identify how well the valuable and expensive resource of cooling capacity is being utilized. A facility’s CCF is calculated by dividing the total running manufacturer’s stated cooling capacity by 110 percent of the IT critical load. The resulting number tells a facility how much opportunity there is by improving cooling management.

Calculating a colo’s CCF is a free, quick, and easy way to determine a facility’s AFM potential. The complimentary CCF calculator is available to all users on upsite.com. Once a site knows this number, it can move on to determining which methods are going to work best, both from an owner’s point of view and the facility as a whole, and from tenants who lease space and are heavily invested in a secure and reliable space for their IT equipment.

Review the application of AFM methodologies

Whether using most or little of their cooling capacity, a colo must still evaluate the best AFM practices available to choose the best options for their space and situation. Following the 4Rs methodology ensures that each level is managed as well as possible before moving on to the next. The 4Rs stand for the Raised floor, Rack, Row, and Room of a facility.

Each “R” can be addressed in turn, depending on how much control the colo may have in the space versus what the tenant wants or expects, and can help guide efforts in monitoring cooling for the entire facility. At the Raised floor level, colos usually have lots of control and can work with sealing openings and “holes,” such as those resulting from cable cutouts or perforated tile placement. At the Rack, they often have little control over directly implementing AFM solutions and can only encourage the use of blanking panels or how servers are loaded. Colos may have some moderate control over the Row, depending on the layout of the room, and at the Room level, a colo has plenty of control, although set points affecting supply temperature may be limited due to SLAs. Again, the SLAs can be an initial and peremptory approach to setting baselines for cooling at the “R” levels that become less available to improving by the colo, once the tenant occupies the space.

Regularly monitor the space

Regularly reviewing the 4Rs and adding cooling management discussions to check-in meetings with clients can ensure that AFM practices are being followed and help to immensely improve on the cooling efficiency of a colo. Hot spots and mixing air wastes cooling and energy, but can be hard to identify. Untended air paths can develop depending on external environments, random obstructions from equipment placed over a vent, blocked tiles, or servers removed from a rack with no blanking panel installed.

Airflow is a finicky thing. A facility might never know where it’ll blow next unless it has both established a strategic plan and continually monitors it to adjust as needed. Of 45 various sites, both large and small, analyzed by Upsite, only 77 percent had placed their perforated tiles in the correct spots. To lower inconsistencies like this, colos need to regularly reevaluate their cooling, recalculate their CCF, and continually assess the 4Rs.

Conclusion

A facility’s cooling capacity should concern both the data center operator and the customer leasing space for their servers. The more efficient the cooling, the greater the profitability and competitive pricing a colo is able to reach. Also, the more capacity and redundancy available in a facility, the better a colo can serve the needs of current customers and open space for future ones. Cooling affects both parties, and AFM knowledgeable and site specific measures ought to be discussed regularly both internally and with customers.

Lars Strong, P.E., thought leader and recognized expert on Data Center Optimization, leads Upsite Technologies’ EnergyLok Cooling Science Services, which originated in 2001 to optimize data center operations. He is a certified U.S. Department of Energy Data Centre Energy Practitioner (DCEP) HVAC Specialist. Lars has, and continues to deliver value added services to domestic and international Fortune 100 companies through the identification and remediation of dilemmas associated with the fluid mechanics and thermodynamics of their data center cooling infrastructure. Lars brings his knowledge and ability to teach the fundamentals of cooling science to numerous U.S. and international, private and public speaking events annually.

[image credit: seeweb: Flickr cc]