Expert Paper – The Eco-Friendly Data Center

Ways and means of improving energy efficiency in IT

Most of the energy used by information and communications systems is consumed by commercial data centers and servers. In Germany, data centers currently account for around 2 percent of total energy consumption. With the increase in mobile computing, rise of social networks, and spread of IT into virtually all areas of private and work life, data volumes are growing exponentially. The result is a continuous increase in both energy density within the typical data center and the resulting cooling requirements. It also means that green IT and efficient, sustainable resource management are becoming key concerns for data centers.

The challenge for data center operators and network managers is to efficiently deploy all data center assets to handle these growing data and energy loads. The solution lies in optimal planning and predictive analysis and prognosis of network, server, load, power, and cooling capacities to ensure maximum sustainability.

Responsible and proportionate use of energy is a vital objective with knock-on effects that help reduce our impact on the environment. A reduction in power consumption also means fewer carbon emissions. The ultimate goal is total carbon neutrality.

According to some estimates, as much as half the energy consumed within the typical data center goes on running the infrastructure, e.g., uninterruptible power supplies (UPSs), power distribution, cooling, etc. The rest is consumed in the actual processing of data. All tools and measures used to optimize this performance come under the heading of “data center infrastructure management (DCIM).”

Before any improvements can be implemented , however, it is essential to have an inventory of existing assets and their respective energy requirements. A typical DCIM system therefore includes an open asset management capability with a comprehensive data model and the ability to collate, evaluate, and present all existing and planned devices within the data center, including cabling, installation details, and all other relationship data. In order to implement optimization measures, it is necessary to be able to access and evaluate the current and planned power consumption data for every device and component.

In order to gather data on the current status of all devices in the data center, the DCIM system must be able to monitor live data from each individual asset and synchronize this data with the data model via standard interfaces or an open API to enable subsequent analysis. FNT Command is a web-based, multilingual, and multi-client software product with a 20-year track record in infrastructure management at numerous major companies. Now with a wealth of added functions for data center optimization, it provides a fully-featured DCIM solution.

Approaches to Data Center Optimization

Current approaches to data center optimization involve introducing advanced energy-efficient technologies (energy-efficient servers, virtualization, more efficient UPSs, alternative cooling systems), consolidating old data centers, upgrading existing data centers, and building new data centers to modern standards using the latest technologies. However, even when using the most up-to-date technologies, there is still considerable scope for improving energy efficiency by optimizing the running of the data center. To ensure maximum reliability, data centers typically require more than a single redundancy option, with most facilities opting for a much greater overprovisioning of power and cooling.

The availability of accurate capacity data and reliable planning information enables optimum scaling of infrastructure components to achieve the ideal balance between service availability (via redundancy and overprovisioning) and energy efficiency. This is made possible in FNT Command by taking a holistic approach to every aspect of the data center facility: power consumption and deployment, thermal loads and cooling capacities, safeguards and redundancies, and the utilization and criticality of systems. All are documented in FNT Command for subsequent analysis. The comprehensive database reveals hidden capacity and enables optimization of processes as well as reduction of overcapacity.

Application Example 1: Energy-Efficient and Fail-Safe Design for Energy Supply

Allocation of power ports for devices in a rack
Allocation of power ports for devices in a rack

Creating resource-efficient and fail-safe IT systems requires effective planning and management of all energy requirements within the data center and facility infrastructures. All devices and consumers must be perfectly matched to the respective power supply network and technologies. To enable this, the FNT Command Power Management module compares the manufacturer’s power consumption data for each device with the measured and empirical values for power consumers. It also provides a full range of planning and management functions. The manufacturer’s power consumption data for individual devices is available from the FNT Command component library, which has information for more than 50,000 different components. It is also possible to use and store empirical and real values obtained from measurement logs. For example, live power consumption values can be imported and analyzed by way of the Data Center Engine module for FNT Command. This data is used for analysis and calculation of threshold values.

Graphical representation of a power distributor
Graphical representation of a power distributor

Using FNT Command Power Management, with its powerful and transparent display, analysis, and planning of the entire power supply system, data center managers have a clear overview of the power supply network and can specify all power consumers with a view to both energy efficiency and the reliability of their IT infrastructure. It is possible to conduct precision analysis of individual areas with regard to both the need for redundancy and the power capacity or fusing provided.

Abb-3_Systemauslastung
Utilization of monitored servers

Application Example 2: Preventing Underutilization of Systems

On average, as little as 10 to 20 percent of server capacity is actually utilized in a typical 24-hour day. In most cases, the available server capacity is based on peak demand, which occurs during limited periods only. One way of improving energy efficiency is therefore to consolidate assets, i.e., to merge multiple, heterogeneous systems in order to reduce the number of servers in the data center. Virtualization is another way of making much better use of capacity. In this case, applications that were formerly spread across multiple computers are grouped together on virtual machines running on high-performance servers. The result is an increase in server utilization by as much as 50 percent. With FNT Command, it is possible to identify underutilized systems and either virtualize them or consolidate VMs from multiple hosts on a single host with optimum utilization. Systems that are no longer required or not being used are shut down automatically.

The comprehensive data model in FNT Command not only stores information about physical servers, it can also provide or import data for VMs and applications. It is therefore easy to identify when monitoring server loads exactly which virtual servers or applications are underutilized.

Application Example 3: Optimum Capacity Planning

Capacity analysis for climate, power, and floor space
Capacity analysis for climate, power, and floor space

If the future requirements and associated capacities of a data center are not fully known and the planning process is correspondingly imprecise, this can often result in precautionary oversizing. Reliable operation is the primary concern, which is why managers tend to err on the side of overcapacity rather than focusing on energy efficiency. Subsequently, however, they come under pressure to identify and consolidate overcapacities. To develop and operate a resource-efficient data center, it is essential to have a detailed overview of current and planned utilization of space, power, and air conditioning systems, as well as the weight of all installed components. This calls for a central control system that shows all information on future and existing resource capacities. In addition, a sophisticated data center infrastructure management (DCIM) solution must support date-specific planning of all assets and components in the data center, including graphically supported planning and analysis of installation space. Only when these conditions are met can a DCIM solution support capacity planning that in turn enables energy- and resource-efficient planning of all data center assets. The FNT Command Data Center Cockpit module offers a comprehensive range of display and evaluation options for all aspects of the data center. It provides an immediate overview of all relevant information regarding power, air conditioning, floor space, and weight. The FNT Command Data Center Cockpit is thus a powerful tool for developing and evaluating multiple data centers. Each one is defined via the rooms assigned. As a result, it is possible to create evaluations of floor space usage, climate conditions, weight and power loads in both graphic and tabular form. It is also possible to manage historization and predict development of the data center on the basis of recorded trends or planned measures. To obtain a visual overview of physical space, “footprint” views of the data center as a whole and of each individual room can be created. Within each footprint, it is possible to display and manage climate zones and physical areas to which devices from the zones are assigned. This view allows detailed evaluations of the distribution of resources for individual groups within the entire data center or in specific rooms.

Analysis of climate capacity in 3D footprint view using FNT Command Aircon
Analysis of climate capacity in 3D footprint view using FNT Command Aircon

Application Example 4: Optimization of Cooling Capacity

Data center cooling is another important aspect of green IT and offers significant potential for optimization. Between 35 and 50 percent of total energy expenditure in a typical data center goes on cooling. It is now possible to significantly reduce that proportion using the latest sensor and cooling technologies in conjunction with a range of physical measures, e.g., hot/cold aisle layout and modular cooling systems.

The ability to plan, analyze, and monitor climate conditions within the data center environment – before, during, and after changes to the IT infrastructure – is essential for ensuring maximum efficiency in the use of data center resources. Cooling circuits and climate zones are planned and managed using the Aircon module, which integrates seamlessly into the central DCIM and asset management platform, FNT Command. The Aircon module for FNT Command combines and documents all functions relating to climate control within the data center. Data centers (multiple rooms), rooms, racks, and logical groups (climate zones) are planned, monitored, and managed with respect to their climatic load and performance. The data center’s cooling circuits and climate devices, as well as the usage of these assets, are documented and monitored with regard to threshold values. Climate control equipment is collectively referred to as climate components, which include climate devices, heat exchangers, and climate generators.
Thanks to the Aircon module for FNT Command, all climate-related assets are available in the central repository.
Choosing the right optimization measures is always dependent on the current status of the IT environment. It is therefore important to conduct a thorough analysis of actual thermal conditions in the data center before making any modification to the climate control system. During that process, various aspects of the IT infrastructure are examined to identify potential for improvement. These aspects include air-flows over and within raised floors, cooling-air losses, cooling system efficiency, and heat generation in the racks.

Hot-spot analysis using CFD in 6SigmaDC
Hot-spot analysis using CFD in 6SigmaDC

In the course of regular energy optimization measures, it is possible to use a DCIM system, such as FNT Command, in conjunction with the latest CFD (computational fluid dynamics) modeling to ensure optimum system cooling even at increased temperatures. To achieve this, FNT has formed a strategic development partnership with Future Facilities, the market-leading provider of CFD solutions for data centers. The integration of FNT Command with 6SigmaDC from Future Facilities, the leading toolset for predictive modeling and simulation of climate control, enables continuous model-based analysis of the impact of changes within the data center on server availability, capacities, and cooling efficiency. Users can simulate and optimize changes in infrastructure before implementation while also identifying unused capacity. Visualization of heat flows and air circulation enables data centers to lower their cooling costs, prevent hot-spots developing, and boost their overall energy efficiency. Thanks to seamless data exchange with the Future Facilities solution, users of FNT Command can leverage their existing asset information for the purpose of predictive analysis, prognosis, and planning of their data centers. After analysis and variant planning in 6SigmaDC, the projected changes can be further refined in FNT Command and implemented with the aid of integrated processes.

Analysis of server utilization in 6SigmaDC from Future Facilities
Analysis of server utilization in 6SigmaDC from Future Facilities

Application Example 5: Dashboards for Decision Making and Monitoring

Special metrics are frequently used to analyze energy efficiency within data centers. Power usage effectiveness (PUE), for example, provides a rough estimate that can be used to calculate infrastructure costs. The PUE is the ratio of the total energy used in the data center to the energy delivered to the computing equipment. In other words, the PUE expresses the efficiency with which energy is used. A PUE of 3, for example, indicates that two-thirds of total energy consumption is lost in heat generation and cooling within the facility, while only one third is used for actual processing. The closer the PUE is to 1, the closer the data center is to maximum energy efficiency. A PUE of 1.3 is excellent and indicates that only around 30 percent of total energy consumption is inefficient.

FNT Command has access to all the relevant data and relationships of each asset within the data center, as well as current and planned energy consumption values. This means that FNT Command can quickly calculate these metrics and present them in dashboards for further analysis and decision making. Thanks to its open data structure, FNT Command has an extremely flexible reporting engine that enables fast database querying and easy calculation of energy-efficiency metrics, such as PUEs and other values.

Application Example 6: Rewarding Colocation Customers for Energy-Saving Measures

Most colocation providers charge their customers a flat rate for power, regardless of whether actual energy consumption by the respective servers reaches the anticipated levels of not. As an alternative to flat rates for the capacity provided, operators can now use a DCIM solution, such as FNT Command, to calculate and charge for actual power consumption based on accurate data from calibrated meters. This power consumption data can be gathered automatically in the FNT Command data model and then enriched with additional information for the relevant colocation customer. As a result, customers have an additional incentive to address the issue of power consumption and, where possible, will opt for more energy-efficient systems, even if capacity is still not fully utilized. All the necessary billing data can be generated from FNT Command in the form of reports.

FNT Command dashboard for monitoring energy efficiency in a data center
FNT Command dashboard for monitoring energy efficiency in a data center

Summary

New technologies and market trends, such as social networks, mobile devices, cloud technologies, and big data, are changing the needs, expectations, and behavior of customers. This in turn is leading to a faster response to market developments and customer demand as well as faster deployment of new or modified IT services. This, of course, requires maximum agility on the part of providers and has a direct impact on the strategic orientation of data centers. Against this backdrop, IT managers are becoming increasingly interested in reliable, efficient, fast, and flexible deployment of demand-based IT services. In order to facilitate the enormous growth in data and provide the services people need, data center operators have to be able to plan their IT and physical infrastructure accurately while having complete transparency across all data center assets.

FNT Command dashboard showing data center PUEs
FNT Command dashboard showing data center PUEs

The challenge for operators and network managers is to efficiently deploy all data center assets to handle these growing data loads. The solution lies in optimal planning, process automation, and predictive analysis and prognosis of network, server, load, power, and cooling capacities. This in turn requires detailed information on the current and planned states of all devices. It is also essential to have scenario capabilities and planning processes that can serve as a basis for accurate prognosis.

In addition to these operational requirements, data center managers face constant cost pressure and have to compensate for the rapid growth in data through continuous improvements in process efficiency. Energy efficiency and environmental sustainability are also a priority – not just in response to legislation, but also as corporate objectives.

Issues, such as risk reduction, the securing of data centers against emergency situations, and adherence to compliance legislation or country-specific audit regulations, require data center operators to have an accurate overview of how all key facility and IT components are linked.

Other issues requiring constant attention include the development of new facilities, the consolidation or replacement of existing data centers, and the need to respond to new technologies and/or architectural changes.

These challenges can be addressed simply and effectively by applying an integrated data center management strategy based on the Data Center Infrastructure Management (DCIM) solution from FNT. The application examples presented above illustrate how a DCIM package offering complete and transparent access to all information and relationships associated with the devices and assets within the data center is an important prerequisite for any optimization measures. The ability to measure, gather, and analyze “as-is” values, combined with end-to-end planning as part of a single data center management tool, provides the basis for achieving greater energy efficiency from data center IT systems.

Leave a Reply

Your email address will not be published. Required fields are marked *