We noticed you're blocking ads.

Keep supporting great journalism by turning off your ad blocker.

Questions about why you are seeing this? Contact us

Font Size

- Aa +

Wed 24 Sep 2008 04:00 AM

Font Size

- Aa +

Energy-smart data centres

Going green can be the secret to significant cost savings as well as aggressive performance growth. Implementing a comprehensive strategy that includes virtualisation and consolidation as well as best practices for power and cooling optimisation can help businesses achieve immediate benefits.

Going green can be the secret to significant cost savings as well as aggressive performance growth. Implementing a comprehensive strategy that includes virtualisation and consolidation as well as best practices for power and cooling optimisation can help businesses achieve immediate benefits.

Companies go green for a variety of reasons. Thanks to ongoing increases in global electricity prices since 2002, green initiatives are often conceived as a response to the rising cost of energy. Some businesses go green simply to improve customer and public perception of their organisation. Still others are driven by environmental concerns to minimise the impact of global carbon dioxide emissions.

Whatever the motivation, IT is a key component of green initiatives. IT infrastructure accounts for a disproportionate share of energy relative to head count and operating costs - and these energy issues can inhibit business innovation and growth.

Existing server room and data centre power distribution systems cannot always support future growth.

Power and cooling capabilities are maxed out in many data centres, and power and cooling expenses frequently outstrip the cost of IT equipment. IT managers are also grappling with operational issues such as difficult-to-manage ‘hot spots' while trying to meet the needs of an ever-increasing server population.

Conflicts between IT and facilities management groups are often the cause of inefficient practices in the data centre. These conflicts typically surface when organisations are planning for upcoming data centre changes such as consolidations and expansions, new re-dundancy requirements, or incremental power requirements.

IT managers are typically concerned with the ability of the IT facility - including its space, power, and cooling capacity - to fully support server consolidation, virtualisation, or high-performance computing initiatives, and the ability of existing server room and data centre cooling systems to support new equipment. For example, in one typical enterprise usage scenario, power distribution and cooling equipment consumed 59 percent of data centre power, whereas compute servers accounted for less than 30 percent.

Existing server room and data centre power distribution systems cannot always support future growth - and organisations frequently cannot afford the costs associated with a major business disruption. Meanwhile, facilities management groups have different priorities when planning data centre changes.

They must consider power usage caps from the utility company as well as energy costs, which often exceed equipment costs. And in many cases, companies are literally running out of space in the data centre because existing server racks are full. Moreover, the data centre may be running out of circuit breakers. Facilities managers often cannot deploy any additional servers until a new data centre is brought online.

Simplification

Because energy consumption is about much more than just the IT equipment, organisations should take a comprehensive approach to greening the data centre. They can start by using energy-efficient platforms with built-in power management features. For example, energy-efficient servers should include high-efficiency power supplies and optimised thermal design, and can also include silicon changes, component changes, and energy-efficient motherboard design.

Organisations can also optimise the office IT environment through effective client power management. In addition, data centre optimisation measures such as enhancing system utilisation through virtualisation and workload management technologies, highly efficient cooling architectures, and energy-efficient storage design can complement these efforts and enhance efficiency.

A positive change in the energy usage of IT equipment can have a direct effect on power and cooling requirements. For example, in one data centre usage scenario, a 10 percent improvement at the server level netted about an 8 percent improvement at the facility level.

However, efficiency improvements in cooling or power delivery were relatively independent; a 10 percent increase in power delivery efficiency manifested itself as about a 4 percent improvement at the facility level.

Pinpointing facility and IT improvements that enable efficient operation can offer significant opportunities. For example, simply running the facility at a slightly higher temperature than usual can often provide an opportunity to increase efficiencies. Coupled with air-handler energy-efficiency options, savings of 5 percent or more at the facility level are not uncommon.

Savings may result from staged compressor operation, slowdown of airflow with variable-frequency drives, increased chiller efficiency, or minimised condensation and re-humidification in the air handler.

Advancing efficiency

Through virtualisation and consolidation, enterprises can reduce the number of physical servers in the data centre while dramatically increasing compute capacity.Virtualised servers can provide more efficient space utilisation, more effective energy consumption in the remaining servers, and lower overall power and cooling requirements than traditional non-virtualised servers.

Virtualisation typically allows organisations to consolidate anywhere from 3 to 20 virtual systems onto a single physical platform. Results vary greatly based on the particular applications, user demand, and configuration of both the old and new servers; organisations commonly report 10:1 consolidation ratios.

Servers designed to facilitate dense computing environments - with substantial memory to help remove barriers to running memory-bound applications - can be the driving force behind a variety of data centre efficiency improvements.

Power and cooling capabilities are maxed out in many data centres.

Managing performance

Replacing legacy servers with higher-performance, higher-capacity systems designed with efficient thermal characteristics and power delivery can be an effective way to cut power consumption and costs in the data centre.

Industry-standard systems specifically config-ured to optimise energy efficiency can help deliver dramatic savings by reducing power consumption through features such as low-voltage processors, high-efficiency memory, high-efficiency power supplies, and energy-optimised BIOS settings.

These servers can deliver significantly greater performance per watt over a similarly configured standard server to help reduce energy expenses.

Today, the power and cooling challenge is focused squarely on servers. However, it is important to proactively manage storage to help prevent potential power and cooling issues.

With data volumes growing exponentially, increasing volumes of storage will likely be required in virtually every organisation. Right-sizing storage systems, drive mixes, and media to information usage can help achieve considerable power savings. Matching application profiles with energy-efficient storage options can help enhance both utilisation and efficiency.

Additionally, because up to 60 percent of the power consumption in a storage array may be attributed to disk drives, reducing the number of physical drives by using centralised storage - such as an Internet SCSI (iSCSI)/Fibre Channel storage area network (SAN) or unified network attached storage (NAS) and SAN data storage system - can also help lower energy costs.

Eliminating boot drives in servers using boot from SAN can yield significant savings for a large fleet of servers. iSCSI helps lower the cost of deploying a SAN by using an Ethernet-based network compared with a Fibre Channel fabric, while single-instance storage helps eliminate duplicate files.

Scaling density effectively

Traditional raised-floor data centres typically have about a 5 kW limit per rack - so, for example, even if 75 percent of the physical space in a rack remains vacant, the 5 kW limit would mean that the rack is operating at its full capacity. Using efficient cooling de-signs can help organisations increase computing density within the same power envelope.

By helping reduce fan power, reduce mixing and short-circuiting of air, and eliminate the negative effects of condensation, alternative cooling products can be an energy-conscious alternative to traditional raised-floor cooling.

For example, cooling systems that based on a pumped refrigerant infrastructure with cooling modules that can be placed directly above or alongside high-density racks can be significantly more efficient than traditional perimeter cooling, helping to overcome cool-ing-related density limitations so that organisations can increase computing capacity in cost-effective increments. Using these systems can help maximise data centre layout and density, allowing organisations to defer the cost of expensive data centre leases or build-outs.

Strategy for growth

Whatever an organisation's motive is for going green, using energy-efficient systems and cost-effective, standards-based power and cooling systems from the desktop to the data centre can help make green goals a reality.

Organisations can extend the benefits of efficient equipment through intelligent data centre design - enabling them to optimise power and cooling configurations to achieve high computing density without performance degradation.  

John Pflueger, Ph.D., manages the Dell technical strategy for facility and system thermals, focusing on ways to improve efficiency. He has 16 years of experience in product development, product marketing, and product management.

Albert Esser, Ph.D., serves as vice president for data centre infrastructure at Dell, where he is responsible for enhancing Dell's enterprise IT solutions by sharing customer insights with the company's server, storage, data centre solutions and services teams.

BEST PRACTICES: Client usage policies that pay off

By reaching beyond the data centre to implement simple usage policies for power management throughout the enterprise, organisa-tions can achieve significant energy conservation and put dollars back into the bottom line immediately. These common sense strategies may seem small, but they can make a big difference:

• Activate power management settings: Utilising power management settings can enable organisations to significantly reduce overall client power consumption.

• Stagger startups: Creating a staggered schedule for powering up computers helps avoid overburdening network servers and triggering increased costs for peak energy demands.

• Turn off desktops at night: During the approximately 100 hours per week that computers are unused, organisations should switch them off to avoid creating heat and increasing demand for power and cooling throughout the building.

Arabian Business: why we're going behind a paywall

For all the latest tech news from the UAE and Gulf countries, follow us on Twitter and Linkedin, like us on Facebook and subscribe to our YouTube page, which is updated daily.
Real news, real analysis and real insight have real value – especially at a time like this. Unlimited access ArabianBusiness.com can be unlocked for as little as $4.75 per month. Click here for more details.