Green Data Centers & Critical Facilities: Aligning Costs ... Center Sustainability 2014--.pdf · 4...

Green Data Centers & Critical Facilities: Aligning Costs, Performance, & Sustainability2013 in Review - Primarily Sourced from December 5th Roundtable

DISRUPTIVE TECHNOLOGY: FEATURED PANELISTS:

2 | Green Data Centers & Critical Facilities: Aligning Costs, Performance, & Sustainability

AGRION | 3

Agrion reports differ from other research out there because the information comes directly from our most valuable resource: our network of informed experts and business leaders. As the industry voice and gatekeeper to our members’ strategies and insights, we like to step out of the way as much as we can so our stable of over 200,000 leading professionals can do the talking.

Our team is charged with curating participants, data, and topics emerging from Agrion roundtables, virtual thought-leadership sessions, taskforce meetings, and summits to ensure that these sectors grow cohesively and our members thrive in a fast-paced, rapidly evolving field. By Agrion’s design, our member-driven approach to research, writing, and projects ensure that we do not overshadow the real industry buzz. We synthesize network contributions into a coherent and (we hope) useful narrative, add relevant external research, and voila – we have a product detailing the conditions of the market, pain points for professionals in energy, cleantech, and sustainability, and some pretty valuable hints about where these industries are headed.

Agrion effectively operates like the hybrid of an industry association and a research firm, our network of professionals driving our agenda and focus while providing a constant flow of content, challenges, and insights as they build lasting connections that foster business collaboration and industry growth. Agrion assembles our members across a variety of onsite and virtual mediums for thought-leadership, idea exchange, and the more than occasional cocktail hour. Out of our assemblies we produce industry intelligence, market analysis, and in some cases larger projects championed by our members.

Agrion initiatives are network driven, and our reports reflect the voice of the industry. Please enjoy the read, and feel free to contact us if you have updated data or knowledge of your own to contribute to our publications and seminars.

Best,

Nicholas J. DavisManaging Director - Agrion Americas

A N O T E F OR T H E R E A D E R :


What is AGRION?AGRION is the global business network for energy, cleantech,

and corporate sustainability. With offices in New York, San

Francisco, Paris, Brussels, and Beijing, and an international

community of more than 200,000 industry professionals,

AGRION provides a platform for members to connect, exchange

ideas, and identify business opportunities on a global scale. Each

office hosts weekly onsite and online conferences to facilitate

face-to-face networking and online collaboration with industry

leaders. Our global network enables AGRION members to

determine critical business drivers, explore innovation, and

realize sustainable growth in a world challenged by resource

constraints, climate change, and evolving political landscapes.

AGRION | 5

Contents

A Note for the Reader

What is AGRION?

Meet the Panel

Industry Insights

References

Abous Us

02

03

06

08

20

23


Bryant Farland

Meet the Panel:Green Data Centers & Critical Facilities: Aligning Costs, Performance, & Sustainability

SkanSka

Senior Vice President, Mission Critical COE leader

Bryant Farland is a Senior Vice President and Mission Critical COE Leader at Skanska, a multinational construction and development company based in Sweden. Skanska is one of the leading ‘green’ construction developers in the world, with a long history of focusing on the sustainability aspects of smart builds. Coming in as general counsel for Infrastructure and Development, Farland has recently assumed this position following the departure of co-panelist Jake Carnemark, now at Skanska technology partner Aligned Energy.

aligned energy

CEO

JakoB

Carnemark

Jakob Carnemark oversees the technology roadmap for Aligned Energy, a pioneering technology developer that creates ultra-efficient advanced solutions in the data center space and the parent company to Inertech & Energy Metrics. Carnemark has over 25 years of experience running mission-critical groups and building large data centers for telecommunications companies and banks.

AGRION | 7

BlaCkroCk

Global Director of Critical Engineeringagrion

Managing Director, Agrion Americas

HerB traCy

niCHolaS

daviS

Herb Tracy is responsible for all of the mechanical, electrical, and plumbing systems in the entire Blackrock portfolio, which includes 100 offices in 70 cities in 30 countries as well as 11 data centers around the world. Fast-growing Blackrock is the largest publicly traded asset management company in the world, with $4.1 trillion under management and another $14 trillion of outsourced analytical solutions, all of which flows through their data centers.

Nicholas J. Davis is Managing Director of the Americas for Agrion, the global business network for energy, cleantech, and sustainability. Agrion mobilizes over 250,000 professionals for intellectual capital exchange, task forces, projects, and research initiatives across verticals such as solar, wind, infrastructure finance, smart grid, microgrids, energy storage, smart cities, and resiliency. Nicholas holds a Bachelor’s Degree from Northwestern and a Master’s in Resource Economics & Environmental Management from Duke University.

amelia

axtell

loCkHeed martin

Energy Advisor

At Lockheed Martin, Amelia Axtell helps facilitate two programs: the Con Edison commercial and industrial program, which gives incentives for building retrofits in Con Ed service territory, and Lockheed Martin’s joint data center program, which offers incentives targeted at data centers for measures that reduce the energy footprint of a data center build or a retrofit, along with 50-50 cost shares for tech assistance studies.


construction developer Skanska, Herb Tracy from the world’s largest asset management group BlackRock, and Amelia Axtell from advanced technologies provider Lockheed Martin. Why is sustainability so critical for data centers, and vice versa? What are the key barriers impeding the industry’s progress today? What are the best opportunities out there for companies to green their data centers, and how can we drive their greater adoption throughout the market?

Introduction

Digital information lies at the foundation of our modern world, but many consumers neglect to consider the growing amounts of data they access and create every day from a sustainability perspective. Storing and processing data requires substantial stores of energy, materials, and water in a world where many of these resources are becoming increasingly stressed and costly to acquire, all of which makes data centers a profound sustainability risk for businesses today. As data production continues to spiral upwards with no end in sight, data centers must find ways to keep up - and for many, this means expansion and even greater demand for land, infrastructure, water, and power, inevitably yielding more waste and higher emissions. For financial institutions, telecoms, technology companies, and other organizations where IT is mission-critical, cost and resource sustainability in data centers is poised to become an ever-growing challenge.

Fortunately, great opportunities to improve efficiency and productivity exist in this space. Even today many data centers are extremely wasteful, consuming vast amounts of energy in inefficient, carbon-emitting ways; in the New York Times’ year-long investigation, some even wasted 90% of the electricity they drew from the grid.1 This presents a clear opportunity to achieve significant savings through the broad landscape of emerging technologies and best practices that are making strides in energy and resource efficiency today. To explore these solutions, we gathered a panel of leading industry experts in our Green Data Centers & Critical Facilities roundtable to share their key, on-the-ground insights and learnings: Jakob Carnemark from disruptive energy solutions provider Aligned Energy (parent company to Inertech & Energy Metrics), Bryant Farland from leading

INDUSTRY INSIGHTS:

State of the Industry

Data production is growing at an unprecedented pace. Even as IT productivity continues to advance dramatically, an insatiable demand for social media, cloud and mobile computing, real-time processing, data analytics, and new features and applications continue to drive the growth of IT systems and their demand for power. Last year over 1.8 trillion gigabytes of digital information was created around the world, according to EMC and the International Data Corporation.1 With prevailing expectations for instant access to information and the increasing centrality of Big Data to an organization’s success, data centers that can meet this demand in the most reliable, secure, and efficient ways are now a business imperative.

The explosion in digital information is driving the expansion of existing data centers, the construction of new ones, and the search for ways to maximize the productivity of existing resources. In the Uptime Institute’s 2013 survey of a thousand data center operators, IT managers, and senior executives, 70% of respondents worldwide have built a new data center or retrofitted an existing one in the past five years, and 36% of data centers are receiving large (>10%) year-over-year budget increases.2 Development is especially strong in the emerging markets of

Latin America and Asia, where 57% and 44% of data centers are receiving large budget expansions, respectively. DCD Intelligence estimates that global investment in data centers grew 22.1% in 2011-12 from $85 to $105 billion, with Asia and Latin America leading the way.3

With 13,000 data centers in the world today - a number expected to double by 2016 - data centers consume approximately 1.2% of all electricity in the world and 2% in the U.S.4 Even as both IT and supporting data center infrastructure have continued to advance in efficiency and productivity with the evolution of technologies and best practices, the sheer demand for data means that data centers’ global power needs doubled from 2007 to 2012, and the average power consumption per rack rose 40%.5 Some forecasts estimate that data centers’ energy usage will comprise the fastest growing component of the ICT ecosystem in the next decade, approaching 1000 TWh worldwide.5 As their need for energy grows, so too will data centers’ carbon emissions and demand for materials and increasingly strained natural resources such as water, barring the ‘greening’ of the industry on a large scale.

Given that avoiding shutdowns has always been a data center’s primary imperative, much of the industry is notoriously risk-averse, wasteful, and slow to change. Hewing to the safe side may be only natural when the average outage costs data centers $7,900 per minute,6 but the market has made great progress in recent years. Newer data centers are substantially more sustainable than older generations and many cost-effective opportunities to retrofit existing sites are available today. While a PUE of approximately 2.0 is the most common estimate of the average data center today, the Uptime Institute’s annual survey of its network (large data center owners and operators) found a decrease in PUE from 2.5 in 2007 to 1.65 in 2013,2

and leading companies in this space are achieving even steeper PUE gains through cutting-edge technology.

AGRION | 9

The Uptime Institute suggests that many of these initial PUE improvements may have been achieved through simple fixes such as preventing bypass airflow (hot-cold aisle arrangement, blanking panels, sealing leakages, etc.) and more efficient UPS and power distribution systems.2 At the margin, advanced solutions providers like Aligned Energy are pushing industry standards forwards. It is these technological advancements that are improving at an extraordinary pace and allowing large, innovative companies to “radically outpace”2 the rest of the industry in adopting leading-edge efficiency best practices. For instance, Blackrock’s flagship Wenatchee data center has a PUE of 1.18 (Tracy), while the new data center developed by Skanska and Inertech for the Canadian telecom Telus yields a PUE of only 1.15 (Carnemark), one of the most impressive in the country.

This ultra-efficiency can drive dramatic cost savings. Entrusted with the $14 trillion worth of assets under Blackrock management, Blackrock’s Wenatchee data center achieves impressive savings by leveraging the region’s inexpensive power (2.5 cents/kW), an evaporative rather than mechanical cooling system for most of the year, and UPS systems that are highly-efficient modular components that can be upsized, downsized, or transported in 25 KW increments (Tracy). Skanska and Inertech’s 2.7 MW Telus data center is 80% more energy efficient7 than traditional sites due to measures such as their pioneering cooling distribution unit that enables free cooling for 98% of the year, virtualization, cloud computing, flywheel storage, and a modular design approach using prefabricated server modules designed for easy, just-in-time expansion.8 Moreover, while the Telus project is sited in Canada, Carnemark explains that the technology is specifically designed to be temperature-agnostic so these PUEs can be achieved in places like Singapore or NYC.

One of the key learnings in our roundtable, however, is that these efficiency solutions are not an exclusive niche too expensive for all

but the largest and most sophisticated companies. As panelist Jakob Carnemark of Inertech’s parent company Aligned Energy explains, the most sustainable solutions are also the best financial decisions for data centers. While branding benefits do exist, greening the data center is also vital for minimizing total costs, mitigating risks, and maximizing productivity - ultimately, optimizing an organization’s bottom line. Despite the higher upfront costs of these more sophisticated measures, paybacks can be very attractive – for example, Telus’s cooling system has a 3-year payback on energy savings alone. Minimal PUEs and reduced maintenance costs drive significant OPEX savings, and Skanska’s modular, “just in time” approach can greatly improve CAPEX and carrying costs8; the Telus data center not only reduced total TCO by 40%, but slashed upfront costs from $200 million to $55 million (Carnemark). As explored throughout

(Above:9 Skanska and Inertech’s eOPTI-TRAX, a highly efficient water cooling distribution unit that is approximately ten times more efficient than traditional chiller plants, even at higher temperatures and lighter, denser loads. Deployed with Inertech’s Rack-Pax, e-Comb, and BeePax, this complete modular cooling solution uses one-tenth the space at one-tenth the cost of running of a traditional cooling system.)

this report, the main issue for the industry lies in the misalignment between supply and demand: while efficiency may make the most economic sense, the market needs new business solutions and greater support and expertise to overcome the CAPEX hurdles, ineffective supply chains, and lack of understanding of the true financial impact of data centers. Moving this market is the goal of incentive programs such as the Con Edison C&I and the NYSERDA data center programs managed by panelist Amelia Axtell’s Lockheed Martin. The Con Ed program provides incentives for building efficiency measures in Con Ed territory, while the joint NYSERDA-Lockheed Martin program targets incentives specifically at data centers, including IT implementation incentives for any measures that reduce the data center’s energy footprint along with 50-50 cost-share incentives for technical assistance studies to explore efficiency opportunities.


Key Issues and Learnings for the Industry

Understanding TCO and the true P&L impact of data centers

“The central problem in the data center industry right now is that companies are doing things on a short-term, cash-balance basis that don’t necessarily make long-term sense from a P&L perspective,” says Jakob Carnemark, CEO of Aligned Energy. “That then drives poor sustainability, when the most efficient P&L impact of data centers also happens to be the most sustainable way to build them.”

As Carnemark explains, when the first data centers were built in the 1980s and early 1990s, the primary cost driver was the IT infrastructure (servers, networking, etc.), which comprised 80% of the data center’s budget. However, while the IT industry has seen an incredible boom in productivity since the 1960s, the construction and energy industries have only climbed upwards in price. Thus, today’s data center costs have shifted from IT (now 25% of the budget) to the supporting infrastructure for that IT gear such as power and cooling systems (75%). A data center’s server is now often said to be less expensive than the cumulative costs of running it over a three-year lifetime.4

These energy needs, an operating expense, represent one of the most significant TCO components in the data center.10 While an under-utilized data center may have a 50-50 OPEX to CAPEX ratio, the TCO for a fully utilized data center is approximately 75% OPEX and 25% CAPEX. “That OPEX model is completely power driven,” Carnemark explains, with 80-85% based on power. Power usage also impacts CAPEX, because reducing energy and improving operating efficiency would decrease utility feed and generator capacity requirements for the same critical load, which would lower upfront capital costs throughout the power chain.10

As a primary and increasing share of TCO, energy costs should be a central concern, but the challenge is that many companies today make their decisions based on constrained CAPEX. As panelist Bryant Farland of Skanska believes, one of the most significant reasons for the success of their Inertech-partnered Telus project was Skanska’s engagement at the earliest stage, becoming a partner consultant rather than a simple contractor. “In the absence of that, Telus would have run its typical procurement methodology – ‘present a CAPEX’,” Farland says. “From our perspective, the construction industry is really a great source of frustration because it is still commoditized in a manner where the procurement methodology is still CAPEX focused; at the end of the day, that’s an easier analysis for them to perform.” Carnemark agrees: “Unfortunately, I rarely see companies that have a unified understanding of the financial impacts of data centers so they can optimize it,” he says. “The pressures on CAPEX are now, ‘let’s go outsource’, when if you look at that model, oftentimes it is three times the cost of what you could do if you could just do it efficiently internally.”

Under-utilization

The reason why companies are CAPEX-constrained is due not only to the challenging economic environment, but the historical track record that data center operators have of overprovisioning. “You don’t want to be in a situation where you build a data center and then your boss comes to you wanting more the next day,” as Carnemark points out, “so people generally want to overprovision a five-year chunk; and then the CFO comes around saying, ‘you’re only 20% utilized, how could you do this?’ So it’s a utilization problem.”

Even as energy costs are becoming an increasingly important component of TCO, data centers remain remarkably

under-utilized. Oftentimes, less than 15% of original source energy is used for the data center’s actual IT equipment.11 In 2010, Viridity Energy’s survey of LexisNexis’s data centers found that three-quarters of their servers used less than 10% of their computational brainpower to process data.1 McKinsey’s yearlong study of 20,000 servers across different industries found that on average, only 6-12% of the electricity was used to perform computations; the rest was used to keep servers idling and ready in case of a surge in activity that could slow or crash their operations.1

Servers are only one under-utilized component of the data center. In Carnemark’s experience, the utilization of UPS systems for major banks is only 30-50%, and the capacity is still insufficient because of the poor fit between the IT equipment and the space design. Moreover, many companies do not understand that both average and peak PUEs must be addressed to actually extract utilization out of their infrastructure. Even as many are making great strides on lowering their average PUEs, they may still not be maximizing their facility’s capabilities because the average PUE drives the OPEX but it is peak PUE that drives CAPEX. “Clients say they have this amazing technology [that will take them] down to 1.6, but their peak is actually 2.0 - and data centers’ costs [depend on] that electrical infrastructure,” Carnemark explains. In New York, most cooling technologies have a 90-120% peak overhead, while Inertech’s cooling infrastructure has only a 10% peak. The reduction in peak load significantly reduces the sizing of the electrical system.

In fact, according to research by American Power Conversion, the single largest cost driver of a data center’s infrastructure TCO is underutilization.12 Estimating a typical TCO of $120k per rack over the

AGRION | 11

data center’s lifetime, they present a breakdown of costs for the typical rack as below:12

To improve TCO, data centers can pursue a variety of strategies, including efficiency improvements, improved planning, rightsizing the system, negotiating costs, and self-service. By far the strategy with the greatest potential benefit is the prevention of over-provisioning, which APC finds can theoretically reduce the cost of infrastructure by around 60%.12

Infrastructure that is never needed never gets deployed, and infrastructure that is needed is only built out when necessary. By deploying a strategy that was not only modular but focused on the right density levels of modular for the best fit, the Telus project could maximize the match between actual IT needs and infrastructure (see below section on Rightsizing).

“It is 100% a utilization issue of buying energy infrastructure on a just-in-time basis for when it’s needed,” says Carnemark. That is why the technology that Aligned Energy develops focuses on two halves: one part involves creating supply chain management systems to connect business processes and IT initiatives with infrastructure and energy solutions to enable just-in-time delivery, and the other half involves delivering these solutions in a much more rapid manner. These halves

are inextricably linked – just-in-time delivery minimizes wasted space and headroom, which means that Inertech’s cooling and power platforms must not only be more efficient but also more modular and rapidly provisionable.

The ideal data center incurs only the infrastructure costs that are actually required at a given time; it uses only the power and cooling infrastructure needed at the moment, takes up only the space needed at the moment, and incurs service costs on only the capital capacity that is actually used. As discussed throughout this report, a wide variety of cost-effective technologies, practices, and deployment strategies now exist that can help companies tackle this under-utilization problem. By bringing together policymakers, construction vendors, and technology solutions providers who are achieving some of the most impressive results in this space, AGRION hopes to create this environment for stakeholders to implement these high-impact solutions at scale.


Approaches to Energy Efficiency in the Data Center

According to a recent report from Lawrence Berkeley National Laboratory, strategies for cooling and IT equipment and load migration can yield a demand savings of 25% at the data center level, with minimal or no impact on operations.13 Each data center will naturally be unique, but implementing the appropriate measures can save an organization millions in energy costs, water needs, space requirements, carbon emissions (and concurrent risk of fines), lost productivity from outages, and vulnerability from future risks.

Rightsizing

According to APC’s research, the greatest TCO savings come from rightsizing data centers such that the infrastructure is fully utilized.12 On a broad level, portfolios can be consolidated to cut waste, achieve scale efficiencies, and direct workload to the most productive sites. For example, after acquiring Barclays Global Investors in late 2009, Blackrock found they now had an “inefficient hodgepodge” (Tracy) of 28 data centers around the world, and embarked on a consolidation and building program to reduce that number to eleven today.

Their goal is to consolidate further to 6-8 data centers, optimizing each.

Capacity planning for a data center can be difficult, especially as the speed of technological progress and data growth can make sites functionally obsolete within a short time. In their consolidation process Blackrock re-examined the reliability needs of each of their applications and concluded that not all required the same resiliency. Instead of designing for zero risk, they decided to support their most critical applications with highly redundant infrastructure and use lower-tier topologies for the more disposable services.14 “Tier IV data centers are phenomenally expensive to build and maintain,” says Herb Tracy, “and what we found during Sandy was that a lot of these tier IV data centers went away. So we don’t build to a tier standard, we build to the resiliency that we require for a specific site – looking at what applications are going to run there, whether they need n+1, 2N, and so on.”

Another strategy to better size data centers is through modular, scalable approaches that enable the faster, just-in-time delivery identified by Carnemark as crucial for tackling

the under-utilization and CAPEX challenges that the industry faces today. Modular designs allow for growth in a pre-planned, stepped fashion - for example, total space and utility capacity can be designed and built upfront, but only individual sections need to be fully outfitted with the UPS and cooling equipment, saving upfront capital and energy costs.10 With phase one of the Telus project completed thus far in two sites (Rimouski and Kamloops), Skanska and Inertech have planned for six phases in total at each site, which would expand the data center from its current 2.7 MW to a combined 32 MW of IT capacity as needed. One analysis by the utility Arizona Public Services found that modular installations can help reduce energy overhead for cooling infrastructure by up to 44%, yielding average annual savings of $200,000 per MW of average IT load along with savings of 1 million gallons of water and 620 metric tons of carbon dioxide.15 The modular approach also enables a safer environment according to Farland, and more efficient systems that resort less often to back-up generators reduce the EH&S risk of emissions fines.

(Left:8 The Telus data center’s architecture uses standardized, prefabricated modules that can be quickly assembled and attached for expansion. The design encompasses not only just-in-time concepts but also ‘just enough’ – multiple tiers are possible, so companies can mix and match to fit their IT needs to the infrastructure.)

AGRION | 13

Modular, scalable approaches are becoming increasingly practical for many components of data centers, including UPS, power distribution, and cooling systems; for example, the Telus design above installs the mechanical and electrical systems in separate prefabricated modules. Other TCO components such as facility improvements, switchgear, and raised floor are more difficult to scale in a modular fashion at this point in time, but APC estimates that half the theoretical savings from rightsizing can be obtained through practical technologies readily available on the market today.12 While the modular data center market is small today (9% of Uptime’s 2013 respondents have deployed modular data centers, with another 8% planning to do so2), Markets and Markets expects it to grow rapidly in the next five years to a $40.4 billion industry, an estimated CAGR of 37%.16 As Farland says, “With Inertech as our technology partner, over the course of the last few years Skanska has really focused on modular deployment both in the retrofit as well as the greenfield space.”

Location

A data center’s location can be a key factor in its power costs, which is why companies such as Google are siting data centers in Finland, where ocean cooling is possible, and why Facebook’s new data center lies just south of the Arctic Circle, where it can achieve PUEs of 1.04 due to free air cooling and slimmed servers.17 In places like Scandinavia, not only is the cool climate ideal but the low energy prices are low and the power grid robust. Other factors that enter into location include the cost of land and construction, the distance and availability of communications providers, the regulatory environment (such as carbon taxes and incentives), labor costs, and latency needs.10

One attendee in our roundtable asked if the entire industry might regionalize in the future, especially as outsourcing becomes more popular; for example, data centers in NYC face 18 cents/kW prices compared to 4 cents/kW

in Virginia. However, our panelists believe that it is not possible for companies to choose sites anywhere they like. Some data centers will be driven by latency concerns - many financial institutions for instance need low latency for major trades, so they will continue to establish data centers in NY and NJ. Another factor is that as analytic data storage platforms such as ebay’s grow to multiple terabytes, the underlying data infrastructure becomes harder to move (Carnemark). With great masses of traffic, “it is much more efficient for companies [like Netflix] to deposit quickly at or [have] multiple accessed instances of infrastructure closer to population densities rather than a central location,” says Carnemark. “So all of these considerations are variables that change the dynamic of where people want to locate data centers.” Future data centers are likely to be smaller and more distributed to maintain performance and move computing and storage closer to end users.18

The broader outlook is that population growth is highest in hotter climates such as Singapore, Hong Kong, and Latin America, where there is limited land, water, and power but a growing demand for data centers. “If you look at the macroeconomic trends,” Carnemark argues, “population densities are increasing in hot crowded places, and the greater the population densities, the greater the need to locate infrastructure at those populations.” This is borne out by the Uptime Institute’s 2013 survey, which shows that Latin American and Asian data centers are overseeing robust budget growth at double the rate of more mature markets.2 Digital Realty’s 2013 survey of North American sites found that NYC, L.A., and London are the top picks for expansion today, with a significant increase in interest in Singapore as well; security, meeting of specs, power costs, and connectivity were the top factors in location choice.19 With the sustained attraction of high-growth urban environments, Aligned Energy is focused on developing technology that is not only water-efficient, but agnostic to temperature and highly dense to minimize space.

IT infrastructure efficiency

According to the Uptime Institute, many simple efficiency opportunities still exist on the IT load side. While around 20% of servers in data centers today are obsolete, outdated, or unused by their estimates, nearly 50% of companies have no scheduled auditing for identifying and removing these wasteful machines.2 In 2011, Microsoft found that average server utilization was still only 15-20%, due to IT departments’ financial constraints, their lack of control of many mission critical applications, and traditional application designs that provide little insight into actual IT resource needs and make it difficult to optimize allocation.20

However, newer generations of servers are evolving greatly. While IT equipment manufacturers in the past may have focused more on maximizing performance, the newest generations of computer hardware also emphasize energy efficiency.21 Because idle servers consume an enormous amount of energy - up to 60% of the power they consume when fully utilized20 - improving server utilization can offer impressive gains in IT energy productivity. One approach to do so is through virtualization software and server consolidation, which cuts the number of individual applications running on individual servers.21 Active power management reduces the power that idling servers draw: research suggests that 1U servers have gone from consuming 60-70% of total power without performing useful work to 25-50% while idle thanks to greater energy awareness.18 Centralized power management and cloud infrastructures are increasingly popular practices that also offer significant efficiency gains. Simply powering down or decommissioning comatose servers would be an ideal starting point, as the Uptime Institute suggests, although the challenge would be the need for staff time.2

Newer servers are not only more directly energy efficient, but also complement other measures to improve the overall facility. By evolving to operate reliably at higher temperatures


(see Cooling section) and being designed for more specific purposes, newer servers can enable efficiency measures throughout the facility. For example, servers that use solid state drives (SSD) can operate at much wider environmental ranges.21 With fewer servers and servers that can operate at higher temperatures, data centers can reduce their cooling needs.

Server racks are also becoming increasingly dense, allowing more computing equipment to fit in the same space. According to Gartner, more than 50% of data centers by 2015 will have a high-density zone, up from fewer than 10% in 2010.22 However, this trend in rising density also presents new challenges. Not all data centers have the cooling infrastructure and design to adopt high-density racks - many older data centers cannot effectively or efficiently cool more than 5 kW per rack, and not all newer data centers can accommodate medium (5-10 kW/rack) or high-density racks (>10 kW) according to Digital Realty.10 The cutting-edge Telus data center has an average density of 14 kW/cabinet, with the ability to support up to 20 kW.7

Facilities infrastructure efficiency

Out of the robust growth in data center investment found by DCD Intelligence, the primary driver has not been IT, but rather the mechanical and electrical sectors (electrical distribution equipment, switchgear, UPS, generators, cooling systems, fire suppression, infrastructure management systems, etc.) along with facilities management.3 The growth in investment from $40 billion to $49 billion in 20123 in such facilities infrastructure underscores the way cost drivers have shifted since the early days of the data center industry in the 1980s. With the expense of this supporting infrastructure spiraling from 20% to 75% of data center budgets today (Carnemark), it is critical to focus attention on these areas.

Introduced by the Green Grid in 2008, PUE is now well-established as the

most common metric for measuring the efficiency of a data center’s facility infrastructure. Leading companies are pushing PUEs closer and closer to one, but more progress throughout the industry is necessary. We discuss below emerging green trends and technologies in the two main categories that comprise facilities’ energy usage - power and cooling22:

Power system efficiency

The power system includes the essential elements for delivering uninterrupted power to the IT equipment, such as the utility transformer, automatic transfer switch, back-up generator, distribution switchgear, UPS, and power distribution system. Failure in any of the components in this power chain would cause a loss of power, which would be disastrous to the data center.22 Thus, redundant system components as well as power paths and bypass systems are necessary. However, not all applications require the most redundant architecture: as discussed in the Rightsizing section earlier, organizations can achieve substantial savings by looking at redundancy on a case-by-case basis for each application, rather than simply designing the entire data center to the highest – and most expensive - tier standard.

The primary driver of the power system’s overall efficiency is the efficiency of the UPS, along with smaller downstream power distribution losses.22 UPS systems are less efficient when operated at low load levels (under 30% of rated load capacity), but this is the common operational range for most 2N or 2(N+1) power systems; older UPS systems may be only 65-75% efficient, wasting the rest as heat.22 Newer UPS units are not only more efficient overall (92-95% at full load) but even more importantly, they can be far more efficient at lower load ranges.22 In their Wenatchee data center, Blackrock looked closely at their equipment options: “Some of the older UPS systems would be 92-94% efficient at 100% load, but you’re running 30-40% load if you’re

lucky, so that means your efficiency’s nowhere,” Tracy explains. “So we looked at products that [can go] down to 15-20% load, but still achieve efficiencies of 96-97.5%.”

The UPS market also faces EH&S issues such as arc flash, which can force a power down to operate a breaker if the potential energy exceeds a worker’s personal protective equipment, although UPS vendors are making progress on this front. According to Tracy, “while switchgear manufacturers have always been very conscious of arc flash protection, the UPS industry is only finally now coming around to worry about arc flash. That’s because there will be staff around the data center who are not electricians or engineers, and we don’t want them walking around [in heavy gear] just because we’re operating electrical equipment.”

Cooling system efficiency

At around two-thirds of the power equation, cooling is the most significant single contributor to the cost of running a data center besides the actual IT load itself.23 In the U.S., cooling costs data centers $2.7 billion annually, a sum that Inertech believes their technology can reduce by $2.5 billion per year. Traditional mechanical cooling systems such as CRAC units and chilled water plants consume enormous amounts of not only power but water, a strategic risk in an increasingly resource- and grid-strained environment, and are often over-sized: in a 2013 study by Uptime, the average running cooling capacity was 3.9 times the IT heat load, which suggests that even simple measures such as turning off cooling units or installing VFDs can yield significant operational savings.24 Retrofitting VFDs for instance can have paybacks of 1-3 years.25

Increasing rack density and new servers’ tolerance for higher temperatures are driving the refinement of a diverse portfolio of more and more efficient cooling

AGRION | 15

technologies, best practices, and optimal airflow designs. As rack densities continue to trend upwards, traditional forced-air cooling methods and raised floor designs are becoming increasingly ineffective,22 but proper airflow designs can cool up to 20-25 kW per rack.18 The trend is for future data centers to adopt a mix of traditional and alternative cooling strategies - for example, a combination of free cooling, evaporative cooling, and a small chiller for the last few degrees.18 In cooler climates, data centers can implement ambient cooling strategies such as air-side economizers (‘free cooling’), or water-side economizers if by bodies of water. Water-side economizers can reduce the costs of a chilled water plant by up to 70%, while mitigating risk because they can provide chilled water even when the chiller is offline.26 Unlike air-side economizers, they can be economically retrofitted, with 1-4 year

(Above: models for hot/cold aisle and cold aisle containment, which reduce the mixing of hot and cold air.27 Below is the cold aisle containment design at the Telus data center developed by Skanska and Inertech.)

paybacks found in California.26

Blackrock’s Wenatchee data center uses indirect evaporative cooling instead of chiller-based air conditioning, and overhead cooling instead of raised floor.14 Simpler, less capital-intensive measures are also widely available - for example, hot/cold aisle containment systems reduce the mixing of hot and cold air and can be combined with other technologies to improve cooling efficiencies even further25; close-coupled or source-of-heat cooling that circulate chilled water to server racks or even processor chips can address high-density hotspots18; proper airflow management such as sealing leakages and installing blanking plates can maximize cooling systems at little cost.25 While many of these concepts are well-known, experts expect continued refinement and broader adoption of these alternative cooling strategies.18


Regardless of the type of cooling system, the amount of cooling required will be reduced if the data center’s temperature can be raised and the range of humidity expanded.21 With no upfront cost, each degree rise in temperature can save 4% in energy costs.25 In 2011, ASHRAE released updated guidelines on temperature and humidity ranges to support the adoption of economizer systems and increase opportunities for data centers to become ‘chillerless’.21 The 2011 guidelines established two new classes of equipment capable of operation at higher temperatures and wider humidity ranges; the highest class (A4) can operate at up to 105 degrees at 8-90% relative humidity.21 Maps published by the Green Grid show that staying within the parameters of the most common equipment class A2 (50-95 degrees, 20-80% RH) would allow 75% of U.S. sites to operate on air-side economizers for 97% of the year, and 99% of European sites to utilize free cooling all year long.21 However, nearly half of the data centers in the Uptime Institute’s 2013 survey reported operating at only 71-75 degrees, unchanged from 2012, although 7% are also now operating their data centers over 75 degrees.2

Of course, other considerations must also be taken into account when raising temperatures. According to Herb Tracy, Blackrock’s data centers

continue to operate at 72 degrees because all the equipment must be in place before temperatures can be raised; having grown through acquisitions, some of their kit can run at higher temperatures, but the older equipment cannot. Moreover, running at higher temperatures may give staff less time to react in case of an event, and the company does not always have 24-7 staffing. Meanwhile, adiabatic cooling or bringing in large amounts of fresh air may create issues in terms of indoor rain, observes Carnemark, which is why Inertech’s cooling platform does not rely on outside air economization. Another concern is the EH&S risk of workers entering hot aisles - at Blackrock’s last data center, raising temperatures to the ASHRAE-suggested 85-90 degrees would require workers to enter a hot aisle of 120 degrees, an OSHA issue. Inertech’s cooling platform has a 4-degree approach from the inlet temperature to the wet bulb and close-couples at the rack out to the atmosphere, which actually eliminates the need for higher temperatures. “On a 78 degree wet-bulb day, which is peak in New Jersey, we can maintain an 82 degree aisle temperature with a 1.5% overhead,” Carnemark says. “The other thing we’ve done a lot of work on is cooling at the rack so you eliminate hot aisles - so not only are we containing the heat, but we’re containing it at the rack and eliminating the hot aisle altogether.”

Inertech’s eOPTI-TRAX, the ultra-efficient cooling distribution unit, can dramatically reduce water and power consumption by replacing the traditional chiller plant and air handling unit with an extended “free cooling” zone.28 While most free cooling modes begin at outside air temperatures under 45 degrees, eOPTI-TRAX’s threshold begins at 85 degrees. This would enable data centers in many parts of the world to operate in free cooling mode approximately 83% of the year, compared with 24% of the year for a traditional chiller plant.28 Moreover, the system’s re-usable refrigerant means that 80% less water is used compared to a traditional open loop chiller. For every 2 MVA of electrical infrastructure, Inertech’s design cuts the power devoted to cooling from 0.7 MW to 0.1 MW, freeing up 0.6 MW of new capacity for IT.

While the technology requires slightly greater upfront investment than a traditional cooling tower, Telus has achieved a 3-year payback on energy savings alone (Carnemark). The water savings from the closed cooling system is especially significant in light of the increasing risks that data centers face from stressed natural resources, congested grids, and storms such as Hurricane Sandy. Current data centers require enormous amounts of power and water to operate, which leaves

AGRION | 17

them vulnerable during resource outages. Carnemark notes that the typical 100-ton unit consumes 4300 gallons of water per day, with 98% of the water going towards the cooling system. As data centers compete with growing populations and standards of living, water usage is becoming a critical strategic issue. “That is why we use this technology - because then water is no longer a source of failure,” Carnemark says. “The system runs more efficiently if you use water, but it will run efficiently without any water at all, and that is important from a risk perspective because you want to eliminate as many points of failure as possible.” Aligned Energy’s technology focuses on a first iteration of 85% water reduction, and a second of 98%.

Sustainability in the data center is thus necessary not only to save money, but also to meet the growing environmental risks of the 21st century. In the footsteps of the now well-established PUE metric, the Green Grid has also created new metrics for water usage effectiveness (WUE) and carbon usage effectiveness (CUE).29 However according to the Uptime Institute, only 21% of data center operators today report carbon emissions, and less than one third track water usage, so greater progress on this front is needed.2

The emergence of DCIM solutions

Tracking these metrics requires more rigorous, comprehensive, and economical data collection and analytics tools. Interest is growing in data center infrastructure management (DCIM) platforms, which allow organizations to monitor their energy consumption and thermals and better address under-utilization issues. From 2011 to 2013, worldwide DCIM revenue increased from $247 million to over $426 million, and is projected to climb to over $690 million by 2016 - a compound annual growth rate of 22.8%.30 In their 2013 survey, the Uptime Institute found that a remarkable 38% of respondents have deployed DCIM software, with Europe leading the way at nearly

50%.2 However, they note that their audience is weighted towards larger and more advanced data centers, and that the meaning of DCIM can be very broad, ranging from “home-grown spreadsheets and monitors” to “an advanced suite of fully integrated tools spanning multiple data centers”.2 Over 60% of respondents in the survey said that cost was the primary barrier to adoption; while small companies are deploying inexpensive DCIM tools (28% spend less than $100k), larger companies are spending significantly more for greater features and scalability (17% spent over $400k). The primary driver for DCIM adoption is to better manage capacity, the key challenge in the data center space identified by our panel.2

The trend for better DCIM tools received a positive reception by our panelists. According to Farland, real-time energy monitoring yields not only savings but underscores Skanska’s performance guarantee. “In all of our projects up in Canada, we had a pretty sophisticated proprietary system built into our building information model,” he says. “It operates very effectively, but it also forged greater confidence for clients that we were really meeting the promises we made when we sold them on this concept . . . So that technology alone has allowed folks to more fully appreciate the efficiencies.” In partnership with another AGRION member, the data infrastructure provider OSISoft, Skanska has developed an intelligent DCIM platform for IT systems and data center facilities, including features for risk analysis.8 Tracy’s Blackrock also monitors down to the PDU or power strip level depending on the application. “We’re trending a lot more to looking at not only what the consumption is, but when it is,” Tracy says, “and we’re making sure that our IT kit is being fully utilized.” The research firm Gartner estimates that DCIM software can help an enterprise reduce consumption by 10-15% on a month-by-month basis.30

One attendee suggests that real-time energy monitoring throughout the day could be applied to peak-shaving.

Jakob Carnemark agrees, pointing out that sophisticated data management platforms can be found if data centers explore outside their industry. While traditional BMS systems that are cumbersome and expensive are still the main technical barrier for data centers, advanced solutions have already been established in other industries. “eBay’s great Digital Efficiency platform . . . is an example of big data infrastructure that’s been used in the mining and oil and gas industries forever coming into the data center and applying the same types of real-time data solutions,” says Carnemark. “Big data optimization of infrastructure is something that has been tackled before and is available out there, so keep in mind there are other technology platforms you can leverage.”

Key Challenges and Barriers

Outsourcing trends

The outsourcing of data centers to third parties such as colocation, cloud computing, and SaaS providers is an ongoing trend that continues to grow, but may present a challenge in sustainability terms. The Uptime Institute’s 2011 survey of very large enterprise data center operators found that 85% employed some form of third-party computing capacity to supplement their portfolio,2 while DCD Intelligence found that global outsourcing rose 31% from $16 to $21 billion through 2012.3 According to our panelists, the fundamental problem with outsourcing is that these third parties are not truly incentivized to become energy-efficient and sustainable. “Colo operators have zero incentive to do anything on efficiency,” Carnemark argues. “They’ll do little things and talk about it but at the end of the day, there is really nothing driving


real wholesale efficiency, whereas if you look at Google and some of these large internet companies, they’ve absolutely got their head around the true costs of data centers.” Herb Tracy agrees: “The fact of the matter is that they are going to pass along their costs to the end user,” he says. “They may be driven by the market because PUE is a big deal, but at the end of the day they are going to spend the minimal amount of money . . . they’re not going to be innovative because it won’t come of out their pockets, it’s going to come out of the end users’.”

While outsourcing may seem attractive for smaller and capital-constrained companies, as Carnemark observes, this is actually due to the sheer inefficiency of data centers and companies’ lack of understanding of their true P&L drivers. Turning to a third party can actually be much more expensive than improving internal efficiency, but many companies choose the short-term, cash-flow-based option because of their experiences with over-provisioning in the past, which has firmed CFO resistance to further investment today. Colocation is an easier expense to understand, and companies may lack the internal expertise and resources to understand the promising opportunities now available out there: “You see a lot of companies that are $5 million per MW P&L impact that are going to outsourcing just because they can’t figure out how to get their internal infrastructure rationalized,” Carnemark says, noting the fact that the largest users have had to abandon the typical supply chain and develop efficient technologies on their own is a sign of a major problem in the industry. While tech giants such as Google or eBay realize the true costs of data centers and have thus advanced their own solutions, the bulk of the market has not caught up. Moreover, outsourcing may not only be more expensive in the long run, but it can also raise critical concerns in privacy, data security, and the lack of exclusive control of one’s future.18

According to Farland, while the

colocation and cloud markets are seeing significant growth, on the enterprise side many continue to look at greenfield development along with a significant emphasis on retrofits and renovations. These align particularly well with the incentive programs that Amelia Axtell oversees at Lockheed Martin. “Skanska sees that as a huge growth opportunity,” Farland says. “It’s a very different procurement methodology, and it’s a different set of people thinking about it in that context - you are really considering it from an operational cost perspective, not necessarily looking just at demand but thinking about things through the context of improving your bottom line via operational efficiencies.” Where Skanska have found the most success is with clients that are truly interested in understanding the TCO, along with ancillary benefits such as brand.

Lack of communications and collaboration

A significant reason for the lack of a unified understanding of data centers’ true financial impact is due to poor communications and collaboration between departments. Nearly 40% of enterprise data center operators do not effectively collect or present cost and performance data to their executives today.2 The power bill should be at the heart of this data, but less than 20% of survey respondents report that their IT organizations pay their data center’s power bill; instead, the vast majority allocate the bill to the facilities or real estate budgets.2

This is the biggest issue that Amelia Axtell faces today as Lockheed Martin tries to support organizations in their efficiency efforts. “The biggest barrier in our program is that the facilities guys, the sustainability guys, and the IT guys aren’t talking to each other,” she says. “Because the facilities guys have the bills, and the IT guys never see the bills or just don’t worry about them because they’re not mission critical, they don’t see how energy efficiency opportunities can really benefit them and the data center.” For example, facility managers may automatically defer to IT because they

are not as knowledgeable about the data center, but the IT managers may not care about the cooling system. “The IT guys are usually stuck to their job day-to-day so reaching them is often tough - we usually have to go through the facility guys, but they’re typically not the ones who necessarily know what’s going on in the data center,” Axtell explains. “So there’s a lot of confusion or backing away from data center projects, and in order for us to really push the program forward, we need to get that conversation and collaboration started between all of those people.”

Carnemark agrees: “You have people who make cooling towers, people who make chillers, and people who make crack units - and nobody wants to change, because they need the other guy or company in the mix to do it. So you really have to start looking at these systems holistically, end to end.” In Farland’s experience from a construction perspective, a wider circle of stakeholders including sustainability managers are in the initial conversations like Axtell hopes, “but at some point that breaks down,” he says, “and ultimately, the procurement comes out in a much more traditional sense and doesn’t account for much of what we’d heard in the earlier discussion, like the significance of operation efficiency or even the TCO.”

Choosing the right partner

Data centers are their own specialty, and should be approached that way. Rather than allowing IT or facilities to design the next data center, businesses should find developers with demonstrated expertise in data center design technologies. However, many new builds continue to use traditional developers who do not fully understand how to deploy the latest cooling measures or streamline airflow design for next-generation systems.18 The challenge of designing, building, and operating high-efficiency, high-density facilities goes beyond the design and engineering of the facility - it requires a well-organized governance structure to oversee the efforts of a specialized,

AGRION | 19

experienced team because mistakes can be very costly.31 In Digital Realty’s 2013 survey of North America, 72% of the respondents planned to use a partner to design, build, and/or lease space.19 The most important factors were demonstrated experience in data center design, operational reliability, and infrastructure management; among other considerations, usage of the latest technology was the most desirable quality.19

Skanska approaches their development work as partner consultants rather than simple contractors, establishing profound engagement at an early stage. According to Farland, partnering from the earliest stage allows them to take a deeply holistic view in fashioning a solution. To mitigate the procurement concerns associated with early and deep partnership engagement, Skanska offer several service options for shared risk; for example, the Telus project was backstopped with a powerful PUE guarantee under any operating conditions for three years.

Conclusion

Sustainability is critical to the core business

Ultimately, improving efficiency simply makes economic sense. The obvious benefit lies in the cost savings, but customer demand for sustainability is also an increasingly salient driver. “We have $14 trillion that come through our DCs through our analytical services, and we’re getting questionnaires with these RFPs before they sign up with us,” says Blackrock’s Herb Tracy. “They want to know what our sustainability policy is, how our efficiency is, how we operate these DCs . . . so aside from taking the cost out of it, we’re being pushed by the business.”

Sustainability in the data center not only converges with financial goals, but also with the core strategic issues such as risk. Efficient data centers are fundamentally more resilient in a

world with increasing environmental concerns and natural resource constraints. This was made clear in the aftermath of Hurricane Sandy: “as more and more of these storms come in, what’s happening is that many of these data centers are out if they don’t have water,” Carnemark says. “Well, if you have a storm like Sandy, how feasible is it to truck water or diesel fuel in? . . . If you build inefficient infrastructure that is highly reliant on your ability to bring in diesel fuel, water, and energy, that is a huge risk issue there.” Aligned Energy’s efficient, renewable technologies thus minimize points of risk. As Carnemark concludes, the sustainability side in data centers is actually profoundly linked to both the financial side and the risk side.

In conclusion, energy-efficient technologies for data centers continue to advance at pace, but challenges in the marketplace impede the widespread adoption of progress and represent strong opportunities for high-impact, innovative solutions. Data centers lie at the nexus of a comprehensive range of needs from all stakeholders, from telecom and financial corporations to society at large. As high energy consumers with a critical need for storage, data centers can play a key role in the success of our future power system. “If you step back and look at the amount of peaker plants and energy storage being built on the grid today,” says Carnemark, “and then look at the way the data center industry is investing in [the same] backup generators and energy storage, the convergence of these two models is an area that yields a huge efficiency not only from a CAPEX perspective, but in terms of building more reliable infrastructure that is better suited for both sides.” In this vision of the future, data centers are no longer energy guzzlers draining the grid, but capacitive infrastructure that actually helps it. This is another promising direction that Carnemark believes data centers will develop in, creating sustainable solutions that benefit all stakeholders. At AGRION, we invite you to follow our future roundtables in this Data Center series for further

in-depth, on-the-ground insight, and contact us if you are interested in local incentive programs such as those facilitated by Lockheed Martin.


References 1. Glanz, James. “Data Centers Waste Vast Amounts of Energy, Belying Industry Im-age - NYTimes.com.” The New York Times. The New York Times, 23 Sept. 2012. Web. 20 Dec. 2013. <http://www.nytimes.com/2012/09/23/technology/data-cen-ters-waste-vast-amounts-of-energy-belying-industry-image.html?pagewanted=1&_r=1&pagewanted=all&>.

2. Uptime Institute. “2013: Data Center Industry Survey.” Uptime Institute - The Global Data Center Authority. Uptime Institute, 8 Aug. 2013. Web. 20 Dec. 2013. <http://c.ymcdn.com/sites/www.data-central.org/resource/collection/BC649AE0-4223-4EDE-92C7-29A659EF0900/uptime-institute-2013-data-center-survey.pdf>.

3. ECM Plus. “Massive 65% rise in data centre power consumption – survey.” ECM Plus - The Voice of Content. ECM Plus, 8 Oct. 2012. Web. 20 Dec. 2013. <http://ecmplus.word-press.com/2012/10/08/massive-65-rise-in-data-centre-power-consumption-survey/>.

4. TriplePundit. “The Sustainable Data Centers of the Future.” Triple Pundit: People, Planet, Profit. RF Code, 21 Feb. 2013. Web. 20 Dec. 2013. <http://www.triplepundit.com/2013/02/sustainable-data-center-future/>.

5. Mills, Mark P. “The Cloud Begins with Coal.” Tech-Pundit.com. Digital Power Group, 10 Dec. 2013. Web. 20 Dec. 2013. <http://www.tech-pundit.com/category/articles/>.

6. The Ponemon Institute. “Data centre outages costing businesses 41% more per minute than in 2010, new survey reveals.” Out-Law.com. Pinsent Masons, 6 Dec. 2013. Web. 20 Dec. 2013. <http://www.out-law.com/en/articles/2013/december/data-centre-outages-cost-ing-businesses-41-more-per-minute-than-in-2010-new-survey-reveals/>.

7. Skanska, and TELUS. “TELUS Intelligent Internet Data Center.” Uptime Institute Sym-posium 2013 Home. Uptime Institute, n.d. Web. 20 Dec. 2013. <http://symposium.uptimein-stitute.com/schedule/1958-telus-intelligent-internet-data-center>.

8. Jerry Bishop. “Skanska Develops Prefabricated Modular Data Center DesignThe Higher Ed CIO.” The Higher Ed CIO. N.p., 2 May 2012. Web. 20 Dec. 2013. <http://blog.thehigh-eredcio.com/2012/03/14/skanska-develops-prefabricated-modular-data-center-design/>.

9. McNevin, Ambrose. “Skanska says its cooling solution is a ‘game changer’.” Datacenter Dynamics. Datacenter Dynamics, 4 May 2012. Web. 20 Dec. 2013. <http://www.datacen-terdynamics.com/focus/archive/2012/05/skanska-says-its-cooling-solution-game-changer>.

10. Julius Neudorfer. “Total Cost of Ownership: An Executive Overview.” Executive Guide Series - Part 2. Data Center Knowledge, 19 June 2012. Web. 20 Dec. 2013. <http://www.da-tacenterknowledge.com/archives/2012/06/19/total-cost-ownership-an-executive-overview/>.

11. Department of Energy. “Federal Energy Management Program: Data Center Energy Consumption Trends.” Office of Energy Efficiency & Renewable Energy | Department of Energy. EERE, n.d. Web. 20 Dec. 2013. <http://www1.eere.energy.gov/femp/program/dc_energy_consumption.html>.

12. American Power Conversion. “Determining Total Cost of Ownership for Data Center and Network Room Infrastructure.” American Power Conversion, 2003. Web. 20 Dec. 2013. <http://www.linuxlabs.com/PDF/Data%20Center%20Cost%20of%20Ownership.pdf>.

13. St. John, Jeff. “Data Centers: IT Efficiency Leads to Smart Grid Savings.” Green Tech-nology | Cleantech and Renewable Energy News and Analysis. Greentech Media, 11 Dec. 2012. Web. 20 Dec. 2013. <http://www.greentechmedia.com/articles/read/data-center-it-ef-ficiency-leads-to-smart-grid-savings>.

AGRION | 21

14. Sverdlik, Yevgeniy. “Blackrock’s infrastructure overhaul.” Datacenter Dynamics. Data-center Dynamics, 29 May 2012. Web. 20 Dec. 2013. <http://www.datacenterdynamics.com/focus/archive/2012/05/blackrocks-infrastructure-overhaul-0>.

15. Clancy, Heather. “Building a better data center, piece by piece.” GreenBiz.com. N.p., 8 Oct. 2013. Web. 20 Dec. 2013. <http://www.greenbiz.com/blog/2013/10/08/building-better-data-center-piece-by-piece>.

16. Botelho, Bridget. “Data center 2.0: The modular data center - How it’s made: The mod-ular data center.” SearchDataCenter.com. IO Data Centers LLC, 6 June 2013. Web. 20 Dec. 2013. <http://searchdatacenter.techtarget.com/photostory/2240185484/How-its-made-The-modular-data-center/1/Data-center-20-The-modular-data-center>.

17. Vance, Ashlee. “Facebook’s New Data Center in Sweden Puts the Heat on Hardware Makers.” Businessweek.com. Businessweek, 4 Oct. 2013. Web. 20 Dec. 2013. <http://www.businessweek.com/articles/2013-10-03/facebooks-new-data-center-in-sweden-puts-the-heat-on-hardware-makers>.

18. Bigelow, Stephen J. “Modern Infrastructure e-zine: The Data Center of the Future is More Science and Less Fiction.” SearchDataCenter.com. Modern Infrastructure, Jan. 2013. Web. 20 Dec. 2013. <http://searchdatacenter.techtarget.com/ezine/Modern-Infrastructure/The-data-center-of-the-future>.

19. Digital Realty. “North American Campos Survey Results.” Data Center Solutions | Dig-ital Realty. Digital Realty, Jan. 2013. Web. 20 Dec. 2013. <http://www.digitalrealty.com/us/knowledge-center-us/?cat=Research>.

20. Aggar, Mark. “The IT Energy Efficiency Imperative.” CDN. Microsoft, June 2011. Web. 20 Dec. 2013. <http://cdn.globalfoundationservices.com/documents/WP_ITEI_Pa-per_052711.pdf>.

21. Neudorfer, Julius. “DCK Executive Guide to Data Center Energy Efficiency.” Data Cen-ter Knowledge. Data Center Knowledge, 14 Aug. 2012. Web. 20 Dec. 2013. <http://www.datacenterknowledge.com/archives/2012/08/14/dck-executive-guide-to-data-center-ener-gy-efficiency/>.

22. ComputerWeekly. “Datacentres will need high-density power distribution within five years.”ComputerWeekly.com. Gartner, 2 Mar. 2011. Web. 20 Dec. 2013. <http://www.computerweekly.com/news/1280095319/Datacentres-will-need-high-density-power-distri-bution-within-five-years>.

23. Homewood, Fred. “As data center cooling costs rise, hot is the new cool: Podcast Q&A.”SearchDataCenter.com. Gnodal, July 2012. Web. 20 Dec. 2013. <http://searchdata-center.techtarget.com/podcast/As-data-center-cooling-costs-rise-hot-is-the-new-cool-Pod-cast-QA>.

24. Kleyman, Bill. “Cooling Capacity Factor (CCF) Reveals Data Center Cost Savings.” Data Center Knowledge. Data Center Knowledge, 19 Sept. 2013. Web. 20 Dec. 2013. <http://www.datacenterknowledge.com/archives/2013/09/19/cooling-capacity-factor-ccf-re-veals-stranded-capacity/>.

25. Carew, John. “Improve Your Existing Data Center Opex.” http://7x24fl-al.org/. 7x24, n.d. Web. 20 Dec. 2013. <http://7x24fl-al.org/Resources/Documents/Improve Opex.pdf>.

26. ENERGY STAR. “Water-Side Economizer.” Home : ENERGY STAR. ENERGY STAR, n.d. Web. 20 Dec. 2013. <http://www.energystar.gov/index.cfm?c=power_mgt.datacenter_efficiency_economizer_waterside>.

References


27. Black Box. “Beat the heat: efficient cooling in the data center.” Inside the Box: The BBOX Blog. Black Box, 27 Apr. 2011. Web. 20 Dec. 2013. <http://bboxblog.wordpress.com/2011/04/27/beat-the-heat-efficient-cooling-in-the-data-center/>.

28. Hilson, Gary. “Inertech, Skanska Partner on Energy-Efficient Cooling Platform.” Net-work Computing. N.p., 6 Aug. 2012. Web. 20 Dec. 2013. <http://www.networkcomputing.com/data-center/inertech-skanska-partner-on-energy-effic/240005042>.

29. Microsoft. “Future Datacenter Sustainability Executive Strategy Brief.” A Day in the Microsoft Cloud. Microsoft, 2011. Web. 20 Dec. 2013.

30. Anderson, Mike. “DCIM software implementation, cloud use on the rise in data cen-ters.”SearchDataCenter.com. SearchDataCenter, 20 Aug. 2013. Web. 20 Dec. 2013. <http://searchdatacenter.techtarget.com/news/2240203867/DCIM-software-implementation-cloud-use-on-the-rise-in-data-centers>.

31. Gunderson, Steven. “Data center design: How to avoid a $5 million mistake.”SearchDa-taCenter.com. N.p., 15 Nov. 2012. Web. 20 Dec. 2013. <http://searchdatacenter.techtarget.com/opinion/Data-center-design-How-to-avoid-a-5-million-mistake>.

References

AGRION | 23

About Us

ABOUT AGRION WEEKLY PROGRAMS

NETWORKING &INDUSTRY COLLABORATIONCOMMUNITIES

OUR OFFICES

AGRION hosts more than 400 events per year across our office locations in New York, San Francisco, Paris, Brussels and Beijing that include webinars to intimate roundtable dis-cussions to large multi-day international conventions. Membership allows companies complimentary access to all events, recordings and reports, and gives them the opportunity to collaborate on a tailored pro-gram to promote new technolo-gies, projects and partnerships. AGRION currently has close to 200 member companies span-ning a wide spectrum of in-dustries and sectors. Through weekly events held online and on-site, AGRION reaches over 200,000 professionals globally.

To find out more about membership, please contact:

[email protected]@agrion.org

Corporate Sustainability

Solar Energy

Green Buildings

Smart Grid

Wind Energy

Smart Cities

Energy Storage

Electric Vehicles

AGRION organizes local and global meetings that span business and technological innovations, market challenges and competition, and policy and financing issues in the energy and sustainability spaces. All meetings are filmed and members have the option to attend meetings on-site (locally) or online. Detailed meeting reports and videos are available following our events.

To view future and past AGRION meetings visit:www.agrion.org

AGRION performs targeted outreach for member compa-nies, ensuring that our unique networking platform will result in high-level industry connections and concrete business leads.

NEW YORK

303 Fifth Avenue, Suite 1105

New York, NY 10016

Tel: +1 212 725 2550

SAN FRANCISCO

5 Third Street, Suite 520

San Francisco CA 94103

Tel: +1 415 882 4615

PARIS

10, rue de Mercoeur

75011 Paris

Tel: +33 1 48 01 08 69

BRUSSELS

Rue du Collège 27

1050 Ixelles, Belgium

Tel: + 33 1 45 23 05 91

BEIJING

Xinyi Commercial Building

N° 3A Xinyijiayuan,

Chongwenmenwai

Chongwen District,

Beijing China 100062

Green Data Centers & Critical Facilities: Aligning Costs ... Center Sustainability 2014--.pdf · 4...

Documents

Transcript of Green Data Centers & Critical Facilities: Aligning Costs ... Center Sustainability 2014--.pdf · 4...