Understanding the tangible benefits
by Kamran Fouladi and Soheil (Sam) Negahbani
How DCIM helps control skyrocketing costs
Data center operating costs have skyrocketed in the past years, with a major contributor being the power bill. Gartner recently estimated that 12 percent of data center operating costs are spent on power, and this cost increases by 20 percent annually. In addition to the higher cost of operations, power poses other major challenges in terms of size, capacity and utilization constraints. The Uptime Institute recently reported that more than a third (36 percent) of data center facilities will run out of space, power and/or cooling in 2011 or 2012. Only a fraction of organizations have the resources to plan and build new data centers, while most will have to resort to strategies such as consolidations, virtualizations, hardware upgrade and other means to meet growing demands and increase the lifeline of their data centers. In recent years, data center infrastructure management has become an essential solution for data centers in their effort to implement these strategies and help deal with their space, power and management challenges.
How DCIM supports the business
DCIM helps data centers identify and determine whether all the hardware assets are needed to meet the business objectives while profiling assets to power usage, cooling requirements, applications running, utilization levels and so on. By providing full analysis of the data center, organizations can report back to management “what if” scenarios on how every move, add or change initiative will affect the status of the data center and properly forecast future improvements to the IT infrastructure.
DCIM enables organizations to systematically control and manage their data centers and help redirect wasted IT budget toward investments that support the core business objectives. Uncontrolled and unmanaged data centers end up wasting the IT budget on counterproductive, time-wasting activities — perpetual firefighting issues.
The ability to visualize, analyze and predict the performance of the data center is among the strongest features of DCIM. This predictive analysis enables the data center operator to examine various and competing concepts before committing to any changes. Moreover, if DCIM is used to analyze the present status and predict the performance of any future changes, then a focus on cooling should be at the epicenter of these efforts.
The challenge of proper cooling management
Lawrence Berkeley National Laboratory has estimated that approximately 50 percent of the power spent to run a data center is spent on cooling the equipment. Additionally, cooling problems have been touted as a major contributing factor to IT capacity limitations, resulting in proposals for major renovations of legacy data centers or even building new data centers. Since data centers undergo frequent changes, it is safe to assume that many of these data centers do not have a clear understanding of how efficient and effective their cooling systems are or whether the design of their cooling systems is still suitable. Ill-conceived operational and infrastructure changes can lead to hot spots and suboptimal equipment layout, while proper cooling would result in reduction of wasted energy and ensure higher reliability and availability. The Environmental Protection Agency estimates up to 30 percent improvement in the energy efficiency of data centers with proper cooling management. This will also extend the life of IT equipment and eliminate premature replacement of critical racks and servers. Understanding and measuring the efficiency of the data center cooling system in any snapshot of time — present or future — is vital for proper operation and modification of the data center.
What CFD can add to the DCIM solution
Computational fluid dynamics (What is CFD?) has become an invaluable tool in helping organizations using DCIM solutions to investigate present and future power performance of data centers, by providing clarity regarding their cooling efficacy. CFD’s capabilities are truly aligned with DCIM’s objectives of modeling, analyzing and predicting due to its ability to create virtual models and simulate the cooling environment of data centers.
- Virtual simulation by CFD provides airflow and thermal patterns of the data center and enables detection of cooling problems (e.g., hotspots, restrictions in the airflow paths, and mixing of cold and hot air streams) for any given scenario or layout configuration. While CFD has long been used by various industries to gain insight into complicated flow problems, the use of CFD by mission-critical facilities has recently begun to gain momentum. Many data centers have been turning to CFD simulation to find ways to improve reliability, availability and cooling effectiveness. In fact, recent surveys indicate that about 50 percent of data centers either have used or are using CFD to deal with their cooling issues.
- CFD facilitates the investigation of cooling effectiveness of any proposed scenario by providing detailed airflow and thermal modeling of the room, aisles and racks while including the most salient physical features of data centers. These features can include equipment layout (IT and cooling), raised floor, dropped ceiling, columns, beams, cables, pipes, tiles (solid and perforated), floor cutouts and more. No other analysis tool can provide the ability to visualize how air — cold supply and hot return — moves through the room to reveal the underlying root causes of cooling problems and determine how much cooling capacity is left at the site.
- CFD case studies performed by Energex Technologies have shown that CFD is also useful in identifying the impact on data center operations from the loss of cooling and in analyzing computer room air conditioner failure scenarios to eliminate uncertainty and validate redundancy. In a recent study, Gartner investigated the usefulness of CFD analysis in data centers and found that data center operators and designers have been able to gain cooling efficiencies and lower operational costs using the results of CFD analysis.
- A lesser-known but unequivocally important benefit of CFD is its unparalleled ability to optimally design or redesign today’s high-density data centers. A typical data center houses various types of equipment with varying thermal and energy characteristics. Such diversity in equipment would generally result in a nonuniform thermal load in the room requiring zoning and other techniques for proper management of the data center cooling operation. The problem with different density zones is that each zone would require different amounts of cooling or even different types of cooling. The optimal design of each zone and equipment layout is possible through trade studies with accurate prediction of the cooling environment for each design alternative.
Energex Technologies believes that CFD is currently the only tool that can provide accurate predictions and show alternative cooling scenarios that provide adequate cooling to the equipment. In fact, to get the full value from their investment, we commonly encourage our clients to go beyond analysis and use CFD for layout optimization.
The benefits of CFD are real
CFD benefits are real, tangible and validated, but data centers contemplating using CFD must move forward with caution. It must be understood that these benefits can be gained only with accurate analysis conducted in the proper context.
- As CFD analyses become more widespread, the cost of these analyses remains a concern to many data centers. CFD analysis does not provide any benefit if it is not accompanied by any physical or operational modifications to the data center. The use of CFD is warranted only if its return on investment (e.g., net savings through reductions in cooling costs) more than offsets the cost of CFD analysis and proposed modifications within one to two years. Managers should conduct a CFD analysis every 12 months or more frequently to keep pace with frequent changes taking place in their data centers. Subsequent analyses should cost considerably less than the initial analysis, by modifying the original model and generating savings on model-building labor costs.
- Data centers must realize that the accuracy of CFD prediction is paramount when undertaking any CFD analysis. CFD analysis that does not contain a sufficient number of simulations or variables can actually be more harmful than not conducting any CFD analysis. A limited number of simulations in an optimization study could lead to less-than-optimized layout or configuration, resulting in oversizing the cooling equipment. An analysis that does not include all the major contributing variables could result in a wrong prediction that could lead to expensive and harmful modifications. For example, an inaccurate prediction could recommend the wrong containment strategy, resulting in equipment starving for cold air, which has been shown to reduce the life span of equipment and jeopardize data center reliability.
The value of CFD analysis in helping data centers reduce cooling costs cannot be understated. Yet, CFD’s ability to increase data center IT capacity and prolong IT equipment lifespan through layout optimization is where CFD can make its biggest contribution to data centers in the future.
What is CFD? CFD is state-of-the-art computer-based simulation predicting what will happen when fluids (e.g., air, water or gases) flow. Using numerical methods and algorithms, CFD provides the ability to visualize and understand the complicated flow phenomena for systems too challenging and expensive to prototype. The factors affecting the path of flow such as objects and boundary conditions are mapped to a three-dimensional computational domain that is subdivided to small and finite volumes. The governing non-linear differential equations of fluid dynamics are then solved in the computational domain using high-performance computers and performing millions of calculations. The solutions to these equations are usually presented in graphical form as flow patterns. CFD analysis for data centers provides the virtual modeling of the room complete with IT and cooling equipment to help visualize and understand complicated airflow and thermal patterns in the room.
Gartner, The Benefits of CFD Analysis in Designing and Running Modern Data Centers, 1 March 2011.
Gartner, Why CFD Analysis Is a Must for Modern Data Centers, 16 March 2011.
Information Week, Data Centers Face Growth Challenges, 20 March 2011.
Lawrence Berkeley National Laboratory, Best Practices for Data Centers — Lessons Learned from Benchmarking 22 Data Centers, 2006.
Soheil “Sam” Negahbani is the CEO and CTO of Energex Technologies. Sam has more than 23 years of experience in building and managing services and technology companies and divisions. His accomplishments include developing and managing storage consulting services for a global systems integration company in North America, founding and managing three successful technology startups, attracting and managing $65 million VC investment, and developing Cisco Systems, first certified global application infrastructure provider solution. Sam has extensive experience in the areas of data center optimization and modernization, data center power and cooling architecture, storage consulting services, network and systems management, corporate standardization processes, and enterprisewide BC/DR development and deployment. Sam is a certified APC Technical Consultant, certified APC Integrated Consultant, DRII Certified Functional Continuity Professional and SNIA Storage Certified Professional. Sam’s educational qualifications include an M.S. in computer information systems (Boston University), a B.S. in information technology (Capella University) and a B.S. in electrical engineering (New York Institute of Technology).
Kamran Fouladi is a computational fluid dynamicist specialized in state-of-the-art airflow, thermal management and energy simulation tools for modeling complex environments and applications. He has more than 20 years of experience as a practitioner, researcher, developer and educator of CFD for aerospace, mechanical and mission-critical applications. Kamran is currently the CFD director at Energex Technologies, and his professional experience includes co-founding two successful engineering services companies and participating in projects of national importance with NASA, Lockheed Martin, United Technologies, Pratt and Whitney, and Orbital Sciences. In the past several years, he has focused on the application of CFD for design and optimization of high-density data centers. Kamran has been teaching fluid dynamics and CFD courses at Villanova University since 2001. He received his Ph.D. in mechanical engineering from Old Dominion University. He is a licensed professional engineer and an active member of several engineering societies (ASME, ASHRAE, AIAA and DVGBC). Kamran has published extensively and has been a frequent lecturer at conferences, technical symposia and workshops. He is a reviewer for journals and technical conferences.