Promises, Promises: Does Your DCIM Solution …...floor model takes years. In addition, if a company...

T H E E N D - T O - E N D R E L I A B I L I T Y F O R U M

FALL 2011 • www.7x24exchange.org

PUBLISHED BY 7X24 EXCHANGE INTERNATIONAL

General Hugh Shelton2011 FALL CONFERENCEKEYNOTE SPEAKER

Power To The People: Be Proactive When MeetingPower Distribution Needs

Promises, Promises:Does Your DCIM Solution Deliver?

PUSHING THE ENVELOPE:Expanding Environmental Conditions In The Data Center

3

WWW.7X24EXCHANGE.ORG

POWER TO THE PEOPLE: BE PROACTIVE WHEN MEETING POWER DISTRIBUTION NEEDS 6

HOW SHOULD I GROW MY DATA CENTER? 12

MAXIMIZING THE RELIABILITY OF STANDBY POWER USED IN MISSION-CRITICAL APPLICATIONS 16

LEVERAGING INNOVATION TO MEET MARKET DEMANDS 26

PUSHING THE ENVELOPE – EXPANDING ENVIRONMENTAL CONDITIONS IN THE DATA CENTER 30

CLOUD COMPUTING AND DISASTER RECOVERY 40

IMPACTS ON DATA CENTERS FROM THE NEW ASHRAE 90.1-2010 ENERGY STANDARD 46

PROMISES, PROMISES: DOES YOUR DCIM SOLUTION DELIVER? 56

HOT AISLE/COLD AISLE SYSTEMS DRIVING CHANGES TO FIRE PROTECTION 60

HOW THE IBC SEISMIC AND WIND-LOADING STANDARDS APPLY TO EMERGENCY STANDBY POWER SYSTEMS 66

INSIDE 7x24 76

Directors and Officers

Chairman of the Board

Robert J. CassilianoBusiness Information Services, Inc.

President

William LeedeckeVanguard

Vice President

David SchirmacherFieldview Solutions

Vendor Representative Director

Juli IerulliCaterpillar

Director

Cyrus J. Izzo, P.E.Syska Hennessy Group

Chapter Representative Director

Michael SitemanJones Lang LaSalle Americas, Inc.

Administrative Director

Kathleen A. Dolci646.486.3818 x103

Membership & Education

Tara Oehlmann, Ed.M.646.486.3818 x104

Conferences

Brandon A. Dolci, CMP646.486.3818 x108

6640

www.7x24exchange.org

626

CONTENTS

4

7X24MAGAZINE FALL 2011

Hope everyone was able to enjoy this summer despite all the catastrophic events!

The hurricanes, floods, fires, and earthquakes of this summer highlight the importance ofhardened Mission Critical Facilities. Design and engineering of Mission Critical Facilities mustconsider these occurrences with regional sensitivity. Resiliency and redundancy are key designcomponents in building a fault tolerant facility. Companies are going to be reviewing the abilityof their data centers to be online all the time even during disaster events. This will provideopportunity for Mission Critical professionals to help design, engineer, build, retro fit, andoperate high availability environments. Projects for hardening facilities will surely become apriority to assist in minimizing risk to business operation. Companies need to be knowledgeableregarding fault tolerant requirements and be prepared to support clients with creative thinkingand expert resources to mitigate risk.

The theme for the 2011 7x24 Exchange Fall Conference being held at the Arizona Biltmore inPhoenix, Arizona November 13 – 16, 2011 is End-to-End Reliability: “Leveraging Innovation”.Conference highlights are as follows:

• Conference Keynote: “Leadership That Leaves a Legacy” presented by General HughShelton, The Fourteenth Chairman of the Joint Chiefs of Staff

• Keynotes by MTechnology and AT&T

• ASHRAE Workshop – Greening of Your Data Center

• Presentations by AOL, Deutsche Bank, and Facebook

• Panel Discussion: The Hunt for Talent – with Panelists from Citigroup, Google, and AOL

• Presentations by The Green Grid and the Uptime Institute

• Exchange Tables on Specific Topics at Monday Breakfast and Tuesday Lunch

• An End-User Exchange Forum Luncheon on Monday

• Vendor Knowledge Exchange on Monday Afternoon

Sponsored Event: “An Evening at Corona Ranch”

The program content is designed to provide value to conference participants and their companiesby focusing on important topics of the day. Energy Efficiency, Cloud Computing, and attractingskilled professionals are highlighted at this year’s Fall event.

I look forward to seeing you at our Fall Conference in Phoenix, Arizona!

Sincerely,

Robert J. Cassiliano

CHAIRMAN’S LETTER

7x24 Exchange Chairman, Bob Cassiliano, presents the keynotespeaker, Robert F. Kennedy, Jr. with a donation to Riverkeeper,NY’s clean water advocate, on his behalf.

6


There’s no doubt about it, the datacenter has become the most valuableasset in today’s competitive businessenvironment and this asset continuesto expand. As Digital Realty Trust hasstated, “The average US enterprisecompany has four data centers inoperation, with an average size of over15,000 square feet (60,000 in total).”Cloud computing is exacerbating thedata center development, as IDCillustrated, “The market for cloud-computing services and software isexpected to grow more than 27%annually over the next five years andreach $73 billion by 2015.”

With the sheer volume of data that isstored, companies are constantlyseeking new and better methods tomaintain, access and retrieve data.But while much attention has focusedon developing the right tools tomanage and store information, there’s

one aspect that’s been lost in theprocess – power distribution.

The bottom line: you can have all theservers in place to drive your datacenter, but all those spinning diskscan present unnecessary costexpenditures if you’re inefficientlyutilizing power to drive these systems.It’s important for companies toconsider power distributionmodels...and take a proactiveapproach to keep costs down andprofit margins up.

It’s All About theInformationTo fully understand the importance ofan effective power distributionstrategy, it’s necessary to take a closerlook at what drives the need for moreeffective management. At the very

heart of this focus is the explosion ofdigital content.

The digital universe is growing at anunprecedented rate. Analyst firm IDCkeeps tabs on digital information andhow it’s impacting the data center.According to the firm’s new 2011report, the volume of information isexpected to more than double everytwo years. This translates to auniverse of information growing nearly50 times by 2020.

“IDC...released its annual survey,which found that the amount of datacreated and replicated is expected totop 1.8 zettabytes, or 1.8 billion TBs, in2011, up from just over 1 zettabyte in2010.”

The survey reports this has majorimplications for anyone who storesand manages information: “Thegrowth of digital data is faster than the

POWER TO THEPEOPLE: Be Proactive WhenMeeting PowerDistribution Needs

by Dave Mulholland

8


growth in storage capacity to storethat data...”

The problem: digital content is puttinga strain on the capacity to deliver it. Ifcompanies don’t have a strategic planin place, they will encounter a majorroadblock – the problem iswidespread. A recent InformationWeek article points to just how seriousthe problem is:

“More than a third...of data centerfacilities will run out of space, power,or cooling, or all of the above in 2011or 2012, according to the UptimeInstitute which recently surveyed 525data center operators and owners…Ofthose that will run out of one or moredata center lifelines, 40% plan to builda new data center, 62% plan toconsolidate servers, and 29% plan tolease colocation space.”

The real issue is companies areworking with data centers that are oldor cannot handle the growth. For thisreason, a proactive developmentstrategy is critical. The articlecontinues:

“A Gartner survey last Novemberfound that 47% of representativesfrom 1,004 large enterprises fromeight countries ranked data growth intheir top three challenges. This wasfollowed by system performance andscalability at 37%, and networkcongestion and connectivityarchitecture at 36%. Sixty-two percentof respondents reported that they willbe investing in data archiving orretirement by the end of 2011 toaddress the data growth challenge.”

Data center managers are squarely onthe spot as their role has become oneof strategic significance. IT is nolonger relegated to the back room:They’re front and center in strategicplanning. IT is a core business unit asthe right projects can accomplish arange of goals, such as growingmarket share, improving cycle timeand introducing the right products tomarket.

“For all this, a CIOs role is crucial, ashe shoulders most of theresponsibility. He is required to workwith heads of other departments tomanage projects...that are businessdriven.”

The question now becomes: How canyou effectively plan for growth?

Getting the Right Plan inPlaceIt’s now clear that information isdriving the need for bigger, faster datacenters. But it’s not just about buildingout, it’s about having the rightstrategic plans in place to ensure datais stored and managed efficiently in acost-effective manner.

It’s important to be a proactive thinker,prior to expansion, as it’s not justadding more servers, but aboutchoosing the proper servers as well asidentifying methods to leverage thetechnology already in place. That’swhy companies need to plan beforethey build. Noted journalist andanalyst Frank Ohlhorst says thesolution is not what it used to be.Budget constraints are forcingcompanies to get creative.

“In the past, when large budgets,venture capital investments andgrowing revenues reigned supreme,solving data center overutilizationproblems was a simple matter ofbuilding a bigger and better datacenter. With today’s economicdownturn, that solution just won’t fly.IT execs must perform due diligencebefore proposing a solution for anover-utilized data center, and thatrequires a long, hard look at theproblem and all available options. Withdata centers, there are many paths toconsider.”

But it goes beyond having the righttechnology. Equally important isstrategic planning for power usageand distribution. All the strategies in

the world mean nothing unless youplan for the expanded use of powerthat’s required to handle the new datacenter requirements.

Powering UpWith the continued growth ofinformation and the correspondingexpansion of the data center, it standsto reason this ups the demand forpower. Recently, the EnvironmentalProtection Agency (EPA) released areport on energy efficiency in the datacenter.

According to the report, powerconsumption more than doubled in alittle more than 5 years, from 2000 to2006. This number was expected todouble yet again by 2011. Thistranslates into big numbers and evenbigger costs. The EPA estimated thatthe federal government alone couldconceivably spend $740 million just fordata center electricity in 2011. Whatwas once a competitive advantage hasnow become a liability.

And it’s not just about costs associatedwith expansion. It’s the dramatic risein the cost of power over recent years.Recently, the US Energy InformationAdministration took a hard look atenergy prices. The Administrationestimates pricing averages 11.6 centsper kWh in 2010 – which is a growthover the cost of 11.5 cents in 2009.This number is expected to hit 11.9cents per kWh in 2011.

This growth has gotten the attention ofthe CFO as increased costs areimpacting the bottom line. ElectronicsCooling points out that power andcooling is even costing more than theIT equipment it supports.

“Historically, the cost of energy andthe cost of the data center power andcooling infrastructure have not beenon the radar for most Chief FinancialOfficers and Chief Information Officersand have not been considered in TCOmodels…This was a reasonable

10


assumption during the 90s, whenserver power and energy costs weresubstantially lower. However…powerdensity has been increasing at analarming rate.”

Thus, an effective power and coolingstrategy is the key to success. CIOsmust keep a mindful eye on the powerthey’re using, while planning for thefuture.

Taking a ProactiveApproachFar too often, companies take areactive approach to powerdistribution. They’re flying blind withthe amount of power they need todrive their systems. In this scenario,it’s a matter of trying, failing andrevisiting before they get the rightcombination. This drives costs up andcan lead to failures in the ITinfrastructure. According to ElectricLight and Power, approximately 92%of all outages result from poorplanning on their power distributionsystem:

“An unscheduled power outage coststhe average large industrialcustomers approximately $40,000,increasing to $75,000 for a four-houroutage. The Electric Power ResearchInstitute estimates annual outagecosts to society amount to $119billion.”

Companies must know their powerparadigm before getting in too deep.They must turn their attention toauditing the data center’s currentpower distribution systems.

A proper audit allows for both IT andfacilities managers to evaluate thedata center’s power path leading tocurrent servers, the efficiency ofexisting transformers and even aKilowatt-per-cabinet analysis ofexisting server rooms. The auditempowers companies to optimize

current power distribution and helpthe systems management team planahead for denser server loads.

It’s important to work with a strategicvendor that understands powerdistribution specifically for datacenters. Rather than writing a specand having electrical contractorsreact, you need to know the currentscenario before jumping in. This alsomeans the facility and IT manager canno longer operate in silos. They bothmust be involved in the power andcapacity decisions.

So the audit is crucial, but what will ittell you? There are many factorsinvolved in a proper audit. You need toknow the number of racks you’ll need,as well as the kilowatts per rack. Youneed to consider both overhead orraised floor configurations.

Determining the RightApproachTraditional thought dictates thattraditional, raised floor configurationwas the only choice for a mission-critical data center in an existing orretrofitted property. But the time ittakes to build out or retrofit a raisedfloor model takes years. In addition, ifa company is renting its data centerspace, the raised floor model makes itdifficult to relocate when new space isneeded.

But there are power distributionsolutions available that allow facilitiesmanagers to circumvent that raisedfloor plan without sacrificingperformance – modular, overhead buslines. These are as easy to install asan overhead lighting fixture systemand do not require further preparationof the concrete floor. On the otherhand, for a facilities manager whoneeds to expand or develop a new datacenter site, a better option would be acontainerized data center that can behoused indoors or outdoors with self-

contained power and cooling options.

But design is not the only successfactor. It’s important to have a fullmonitoring system in place as well.System improvements simply cannotbe made without taking the right DCiEand PUE energy and power usageeffectiveness measurements. Thesemeasurements enable companies todeal with higher energy costs, theneed for increased data centercapacity, and the increase in energyusage.

An effective Branch Circuit MonitoringSystem brings the ability to monitordown to the server branch and circuitbreaker level. It provides real-timeload currents and voltages. Thecorrect calculations will focus on totalpower on the panel level (KW, KVA,KVAR, PF and load percentage) – aswell as the branch level. The rightsystem will also provide warning andalarms alerting customers to theiradjustable current, and whether acustomer is over voltage or undervoltage.

Harnessing the PowerIt’s not just about buying and buildinganymore when it comes to today’s datacenters. Information explosion isdriving the build-out of larger, energyconsuming data centers. Effectivepower management has taken centerstage in this discussion, and the rightstrategy could save a companymillions.

Companies must take a highlyproactive approach to power planning.They need to have an accurateassessment of current usage and whatit will take to plan for the future. Intoday’s world, planning is everything,and you need to partner with a vendorthat could help you power the datacenter without breaking the bank –Power to the People!

Dave Mulholland is Vice President of Marketing, Service for Power Distribution, Inc. (PDI). He can be reached at [email protected]

Knowledgeable data center ownersand operators have learned, to theirregret, over many years that buildingtoo much or too little data centercritical IT load capacity (simply calledcapacity in this article) at one timeoften causes major problems. Thisarticle examines the typical problemsencountered, proposes a strategy forrational growth, and illustrates anexample of how to implement thestrategy.

When too much capacity is added atone time, three undesirable thingsinevitably happen:

• The Owner makes a larger CapitalInvestment (CAPEX) than isnecessary, wasting limited funds tobuild capacity that will not be neededfor many years. The Owner cannotmake a suitable Return onInvestment (ROI) until the overbuiltcapacity can be put to beneficial use.For example, if the Owner’s ROIcriterion is five years payback, thenit makes no financial sense to build

capacity that will first be used morethan five years in the future.

• The data center operatesinefficiently because only a smallpart of the installed capacity is used,driving up the Operating Expense(OPEX), Total Cost of Ownership(TCO), and PUE. Data centers aretypically designed to operate mostefficiently at full capacity, and theefficiency drops off dramatically atpartial load. For example, a typicallow voltage, double conversion,static UPS system operates at 77%efficiency while at 10% load, and92% efficiency while at 50% load orgreater. A Dynamic Rotary UPS(DRUPS) has essentially constantlosses, so a DRUPS that is 96%efficient at full load will be 71%efficient at 10% load. A facility PUEforecasted to be 1.5 at full load willdeteriorate to 1.8 or more whenoperated at half load.

• The data center is more difficult fortechnicians to operate at low load.

Water-cooled centrifugal waterchillers have difficulty operating witha low percentage load and frequentlysurge and even drop off line. UPSbattery service life often shortens ifthe UPS system is operated at lowload. Variable speed motor drivesbecome cranky and less reliable.Generator sets smoke and foul theengine exhaust system withunburned carbon, which is calledwet stacking.

Conversely, when too little capacity isadded at one time, at least threeundesirable things can happen:

• The cost per kW for added capacityis greater for small increases thanfor large increases. The economiesof scale and reduced costs expectedwhen adding 2mW of capacity don’thold true when only 600kW ofcapacity is added.

• Getting the electrical and coolingcapacity increases to line up in asimple, logical manner becomes aproblem. If UPS capacity is added in600kW increments, then thecorresponding cooling loadincrements are about 200 tons. If thechillers are 1000 tons each, thenanother chiller, cooling tower,chilled water pump and condenserpump might be needed for a 200kWUPS capacity increase, or adding200kW cooling capacity incrementsmight require less efficient systems(such as air-cooled chillers) thanmore efficient technology with largerincrements (such as high efficiencywater-cooled centrifugal chillers).

• The risk to the operating IT loadincreases every time capacity isadded to the data center. The moreoften we add capacity, the more riskwe incur. Regardless of every effortto eliminate human error throughelaborate planning and Methods ofProcedure, we court disaster everytime we add capacity in a live datacenter. This risk is particularly acutewhen capacity must be added to aUPS system that is already carryingcritical IT load. Don’t forget that theUPS system must be taken off line toadd the additional module and toperform testing and commissioning.

12


by Christopher M. Johnston, PE

14


So, how should we establish a rationalstrategy for planning load growth inboth capacity and frequency? Whiledetails vary with different projects anddifferent owners, there are four majorcomponents to the strategy: (1) forecast load growth, (2) determineDay 1 capacity, (3) determine thetiming of future capacity additions,and (4) master plan the ultimatefacility.

The first step is to prepare a forecastof capacity needed over the operatinglife of the facility. Begin with the initialload when the data center opens (Day1) and end 15 or 20 years in the future.Those preparing the forecast shouldn’tworry that it won’t be perfect (unlessyou have a better crystal ball thanmine), but it will be the best initialeffort. The forecast should be kept up-to-date during the life of the facility soit serves a guide for timing capacityadditions and updating the strategy.For example, if the next capacityaddition is required when the criticalIT load reaches 3mW, and the updatedforecast is that the load will reach3mW in two years, then the capacityaddition project should be begun farenough in advance so it is tested andcommissioned 2–3 months before thecapacity is needed.

When the data center opens on Day 1,you should have enough capacity forthe first four or five years of operation.This preparation will permit thetechnical staff to become very familiarwith the facility and all operations andmaintenance modes before the firstcapacity addition is made. Addingcapacity when the data center is newand the operations staff is not well-versed and comfortable in alloperating and maintenance modes is arecipe for disaster. It’s far better tobuild enough capacity for four to fiveyears so all of the inevitable bugs willbe revealed and resolved beforeadding capacity in a live data center.

You should plan for phased capacity

additions at intervals of more than twoyears. This approach is sometimescalled the “modular” concept of datacenter design. Additions at two yearintervals or less keeps the data centerin a state of constant construction andcommissioning, thus increasing therisk to the operating critical IT load.Hard experience has taught thatpoorly planned construction causesmore outages than routine operationin a live data center.

Master planning the facility for allphased capacity additions can preventfuture problems by (1) identifying risksto the operating critical IT load, andthen (2) establishing strategies tomitigate the risks. In a perfect world,each capacity addition would be freestanding and not connect to anyexisting systems. In the real world,such a strategy produces very costlyand inefficient designs. Expanding thecapacity of existing systems is oftenmore expedient, such as adding a 1000ton chiller to an existing 3000 toncentral plant rather than building anew central plant with two 1000 tonchillers. Master planning shouldidentify how the addition will beconnected and should provide propervalves, circuit breakers, andconnections points to facilitate theaddition at minimal risk. And whileyou’re doing it, don’t forget to providesome additional space for futureinfrastructure upgrades.

Here’s an example of how this strategywas implemented in the conceptualdesign of a large data center.

1The Owner’s IT team prepared acritical load growth forecastshowing 3mW of critical load at

Day 1 initial operation, increasing to10mW in the fourth year, and 40mW inthe 15th year. The forecast wasproduced with some trepidation butwas necessary for not only this effort,but also for use in Total Cost ofOwnership (TCO) calculations used toassess alternatives.

2Using the forecast, the teamdetermined that the initialcapacity would be 10mW,

adequate to meet the forecast throughthe fourth year.

3 Again using the forecast, itseemed logical to design andconstruct the facility in four

phases of 10mW each. Each phase willbe built in a separate building on thesame site, with Phase 1 adjacent toPhase 2, and Phase 3 adjacent toPhase 4. At the same time that Phase1 is completed, the building shell forPhase 2 will also be completed. WhenPhase 2 is completed, the buildingshell for Phase 3 will be completed.This procedure will continue forPhases 3 and 4. A similar design couldbe easily done for a multi-story datacenter, except that the entire core andshell would be constructed as part ofPhase 1.

4 As an example of what’s doneduring master planning, anelectrical utility substation will

be installed on the site with 40mVA ofinitial capacity and capable ofexpansion to 80mVA capacity withoutaffecting the operating critical load.Spare circuit breakers will be providedin the switchgear and cables will berun out to manholes for extension tofuture phases without having to workinside energized electrical switchgear.All of the utilities and spaces areplanned in a similar manner. It’s notessential to plan down to the wallswitch and receptacle level, but it isessential to plan major systems so youdon’t find a fatal flaw in the originalplan as you begin design for Phase 2.

In this article we have examined at ahigh level the problems encounteredin adding capacity, proposed a strategyfor rational growth, and illustrated anexample of how to implement thestrategy. We hope that this techniquewill assist the industry in determiningwhen and how to add data centercapacity.

Christopher M. Johnston, PE is Critical Facilities Chief Engineer for Syska Hennessy Group. He can be reached at [email protected]

16

Mankato, MN – While standby powersystem reliability is a concern for anyfacility, it is especially important formission-critical applications such ashospitals, data centers,telecommunications, government,municipal water and water treatment.Additionally, there are numerousorganizations that rely on standbypower systems for business continuityand to reduce exposure to monetaryloss resulting from a utility outage.

To maximize reliability, facilitymanagers need to understand andconsider the critical factors that gointo specifying, installing andmaintaining a standby power system.These factors can be grouped into fivecategories:

1. Generator set design andmanufacturing quality

2. Generator set sizing and powersystem design

3. Commissioning and operatortraining

4. Maintenance and periodic testing

5. Code compliance

While no mechanical system can beexpected to perform with 100 percentreliability over time, modern dieseland spark-ignited standby powersystems come very close to this ideal– provided they are properly designedand maintained. In fact, power systemcomponent failure is a fairly rareevent, whereas the vast majority ofproblems result from human error orneglect. This paper will examine thefactors that contribute to powersystem reliability and suggest ways tomaintain it at the highest possiblelevel.

What is “reliability”?Before discussing ways to ensurebetter power system reliability, it isimportant to define the term. TheInstitute of Electrical and ElectronicsEngineers’ (IEEE) Reliability Societydefines reliability this way:

To a great extent, reliability can bedesigned into generator sets, transferswitches, switchgear and controlsystems to increase the likelihood thatthey function as intended. Of course,the other part of the definition relatesto maintenance, testing and support –all human activities that must becarried out as part of an overall planto maximize reliability.

Another way to look at reliability is toconsider it from an economic point ofview. In general, to get the highestreliability, facilities will incur greater


By Michael Dauffenbach

MAXIMIZING THE RELIABILITY OF STANDBY POWER USED IN MISSION-CRITICAL APPLICATIONSIdentifying equipment, systems design and maintenance procedures thatcontribute to dependable Emergency Power Systems

Reliability is a design engineeringdiscipline which applies scientificknowledge to assure a product willperform its intended function forthe required duration within agiven environment. This includesdesigning in the ability tomaintain, test, and support theproduct throughout its total lifecycle. Reliability is best describedas product performance over time.

18

costs for redundant equipment,advanced system design and morefrequent maintenance.

For organizations that face life-safetyrisks or severe financial losses if theirstandby power system fails, it is oftenprudent to invest more to attain thehighest possible measure of reliability.For example, this often meansdesigning for N+1 redundancy in utilityfeeds, generator sets and UPSsystems as recommended in theUptime Institute’s Tier IV designtopology. While this redundant systemdesign approach comes at a higherfirst-cost, power reliability andavailability improve. N+1 redundancyalso enables periodic equipmentmaintenance to be carried out withoutaffecting the availability of the standbypower system.

Actual measured availability of powersystems in mission-critical datacenter applications ranged from 99.67percent to more than 99.99 percent ina 2006 study by the Uptime Institute.At the higher end of the availabilitywere systems with N+1 redundancy.However, the Uptime Institute noted inits study that actual availability wasbelow the vaunted “Five Nines”(99.999 percent) sought by manymission-critical applications.

However, this higher cost must beweighed against the cost of powerinterruptions that disruptmanufacturing or business. Industrystudies have found that the cost ofdowntime for a major corporation canrange up to $6.5 million per hour. Forcertain businesses, it is clear that theadditional investment in a morereliable power system will be a wisedecision. In addition to financialconsiderations, the ability to maintainelectric power to systems whose lossmay impact human safety, such asventilation systems, elevators andstairwell lighting, is also critical.

Each organization has to determinethe level of reliability it can afford, or,conversely, the amount of risk it cantolerate. And, while spending moremoney for redundancy to eliminatesingle points of failure generallyincreases reliability, it also increases

complexity, which at some point may,itself, threaten reliability. Afterdetermining what level of reliabilitymay be acceptable and affordable, anorganization must turn to the selectionof equipment and suppliers.

1. Generator set design Engines – Diesel engines are some ofthe most reliable prime movers everdesigned and are the most popularchoice for standby power applications.For optimum reliability, look forengines that are designed specificallyfor power generation applications andnot simply adapted from off-roadheavy-equipment applications.Engines specifically designed to powergenerator sets have been optimized tostart and assume full load in 10seconds or less and run at a constantrpm (1,500 rpm or 1,800 rpm).Because they operate at a constantspeed, generator set engines alsohave different turbochargers thantypical off-road or on-road engines,have different combustion parametersand need to meet different emissionslevels.

For the highest reliability, look forgenerator sets with engines that havesome measure of reserve horsepowercapacity at the alternator’s nameplatekW rating and a low brake meaneffective pressure (BMEP). ISO 8528-5identifies larger engine displacementand lower BMEP as key factors in agenerator set’s ability to accept loadwithout an undue drop in outputvoltage and frequency. Enginemanufacturers vary in their approachto this issue. Therefore, when one-step load-acceptance is called for inmission-critical applications, select amanufacturer that can provide agenerator-drive engine with thehighest displacement and lowestBMEP relative to nameplate kW rating.

New engine manufacturing qualitystandards practiced by somecompanies have helped increase themean time between failures (MTBF) onengine components by a significantfactor. Manufacturing improvementshave included significantly highermachining tolerances, better

metallurgy, sophisticated qualitycontrol systems (ISO 9001: 2008) andimproved inspection and testing. Inaddition, the best modern engines arecomputer-controlled – which not onlyimproves performance, economy andreliability, but also limits thepossibility that an operator mayinadvertently alter the engine’sperformance characteristics. Each ofthese incremental design andmanufacturing steps taken by severalleading engine companies helps toassure power system operators thatmechanical failure of the prime moverwill be a very unlikely event.

Alternators – As a major componentin the standby power system, theability of the alternator to supply itsrated kVA and resist damage fromtransients is crucial to the reliability ofany power system. While most majormanufacturers utilize standardalternator protection schemes, morerecent microprocessor-based controlstake transient protection to a higherlevel. These introduce the feature ofprogrammability into protectivedevices for over-current protection.For example, with modern molded-case circuit breakers (MCCB), thesystem designer can set the devices toactivate very near the protection limitsfor the alternator. Older analog fault-protection methods had a lot of grayarea, meaning that the protectionpoints had to be quite conservative,leading to more fault occurrences thanreally necessary. The reliability ofolder thermal-magnetic breakersdepended on the amount of regularexercise they received.

The type of alternator selecteddepends not only on the size of theelectrical load it must supply, but alsothe types of loads. Factors to considerwhen specifying alternators for themost reliable power systems includetemperature rise, fault tolerance andreactance issues, especially withlarge, nonlinear loads such as UPSsystems and large motors.

2. Generator set sizingand system design Appropriately sizing a generator set


20

for the specific application has a majorimpact on power system reliability.Some generator sets that are requiredto pick up a load equal or close to theirnameplate rating may not perform asintended. While the generator set maystart and run, it may not be able toassume the facility load in one step asrequired by NFPA 110, or it may takelonger than the required 10 secondsfor mission-critical or life-safetyapplications. Unless all critical loadsare properly supplied within the 10seconds as required by NFPA 110, thestandby power system cannot beconsidered to be reliable for mission-critical applications. Consult theengine/generator set manufacturerduring the planning stages to be surethe generator set will be capable ofproviding the expected transient loadperformance.

System design

Design considerations such as N+1generator set redundancy, transferswitch selection, controls and ambientconditions play an enormous role inmaximizing reliability.

• N+1 system design – The UptimeInstitute has developed a TierClassification of I – IV to describe thedesign topology of standby powersystems used in mission-criticaldata center applications. Tier I

topology (see Figure 1) represents apower system design with noredundancy – typical of mostcommercial standby powerinstallations. In practice, accordingto the Uptime Institute, this designscheme results in approximately99.67 percent availability annually.

Figure 2 shows a Tier IV topology thatis recommended for mission-criticaldata center applications with thegreatest need for power availability.With N+1 redundancy in utility feeds,standby generators and UPS systems,such a system is expected to deliverannual availability of approximately99.99 percent. A standby system withmultiple generator sets (eitherparalleled or segregated by loads)improves reliability because thescheme increases the likelihood thatat least most of the generator sets willstart and run as intended. In aparalleled N+1 system design,typically all generator sets start whenthere is an interruption in utilityservice. With proper configuration ofthe switchgear, the “extra” generatorset will shut down after a time if allthe other generator sets start and runnormally.

• Transfer switches – The selection ofthe transfer switch depends on thetypes of loads on the system.Choosing the right mode of

operation (open, closed orprogrammed) for the application cango a long way to minimize the stressof load acceptance on the generatorset. This is especially true infacilities with large motor loads orlarge nonlinear loads such as a UPSsystem, motors with variable-speedcontrol or other electronic loads.

• Control systems – Controls havebeen among the fastest-evolvingpower system components. Bothanalog systems andmicroprocessor-based digitalsystems offer high reliability, andboth continue to be manufacturedand used, depending on theapplication. There is a goodargument that the monitoringcapability of digital systemsenhances reliability of the totalsystem by helping to identify issuesbefore they become problems.

Power systems that feature theflexibility inherent in open-protocolcontrol systems and software ensurebetter compatibility and systemintegration – which leads to increasedreliability. While certain proprietarycontrol protocols may exhibitacceptable reliability as a stand-alonesystem, the likelihood of failureincreases as these systems areinterfaced with components fromother manufacturers or software from


Figure 1. A typical Tier I design topology for astandby power system serving a few criticalloads. Such a system has been shown to exhibitabout 99.67% annual availability.

Figure 2. A standby power system with Tier IV design topology and full N+1 redundancy inutility supply, UPS systems and generator sets. This design has been shown to exhibitupwards of 99.99% annual availability.

22

third-party suppliers. Proprietarycontrol systems also complicatetesting and maintenance if there arecompatibility issues betweencomponents and subsystems.

• Ambient conditions – The operatingenvironment must be taken intoconsideration when designing andinstalling a standby power system.Power systems in coastal regionsare likely to need more frequentmaintenance and inspection due tosalt air. In areas of the earthquake-prone western United States, powersystems used for mission-criticalapplications need to be designed andbuilt to meet the seismic standardsof the International Building Code(IBC). Similarly, site altitude andtemperatures are important factorsin system specification and designthat may affect generator set rating.

3. Commissioning andoperator trainingProper commissioning is essential tothe startup of a standby power systemand ultimately is essential for thesystem’s reliability, regardless of itssize, type or industry. As powersystems become more complex, thecommissioning process becomes evenmore important to confirm that theentire system functions as designed.

The purpose of commissioning is toverify that all components in thepower system are functioning asdesigned in the event of a poweroutage. In fact, it is duringcommissioning that most design orinstallation flaws are uncovered. Thegenerator set must start and acceptload, and all alarm functions need tobe tested and verified. If the systemdoes not function as required, thenremedial measures need to be taken.Following a commissioning protocolsuch as ASHRAE 0-2005 and themanufacturer’s guidelines will ensurethat the commissioning process willbe implemented in a coordinatedmanner.

The commissioning process is also aneducational opportunity for systemoperators and maintenancepersonnel, and it sets a baseline for

future operational analysis. Making avideo of the initial training session isone way to help new personnel quicklyadapt to the established operating andmaintenance routine.

Proper training of operating personnelis essential for a reliable standbypower system since human error orneglect is responsible for the majorityof power system failures. Personneltraining begins during thecommissioning process and shouldcover system operation, record-keeping and periodic maintenance.Operators must be familiar with all thepower system components, alarmconditions, operation and maintenanceprocedures. Special attention shouldbe given to critical subsystems suchas fuel storage and delivery, startingbatteries, engine coolant heaters, andair flow in and out of the generatorbuilding or enclosure. Frequentretraining is also necessary, alongwith making sure that personnelmaintain an operational history of thepower system. Consult your generatorset manufacturer about factorytraining opportunities available tocustomers.

4. Maintenance andtestingOnce a power system has beenproperly designed and commissioned,the most important factor in its long-term reliability is regular maintenanceand system exercise. Someorganizations undertake themaintenance themselves, while othersopt for maintenance services directfrom the generator set manufactureror its distributor. See Figure 3.

Preventive maintenance of generatorsets should include the followingoperations:

• Inspections

• Oil changes

• Cooling system service

• Fuel system service

• Testing starting batteries

• Regular engine exercise under load

It is important to establish amaintenance schedule that is basedon the specific power application andthe severity of the environment. Forexample, if the generator set islocated in an extremely cold or hotclimate, or is exposed to salt air, thegenerator set’s manufacturer can helpdevelop appropriate measures to dealwith these special needs.

Like regular maintenance, periodictesting is required by code in mission-critical applications. It is best toexercise a generator set under theactual facility load it will be expectedto supply in emergency conditions.When operated with the actualbuilding load, the entire power systemis tested – including the automatictransfer switches and switchgear.

Operating a generator set under no-load conditions can adversely affect itslong-term reliability if the generatorcannot get up to an exhausttemperature of approximately 650degrees F before the test is over. It isvery important that both the engineand generator reach this minimumoperating temperature in order todrive off any accumulated moisturethat may have condensed in thesystem. Under heavy load, dieselengines come up to operatingtemperature in a matter of minutes,


Figure 3. Regular exercise and maintenance ofthe complete power system are very importantfactors in high reliability.

24

whereas, without load, they may notreach operating temperature evenafter prolonged operation.

Most manufacturers recommend thatgenerator sets be exercisedperiodically, loaded to at least 30percent of rated capacity. If it is notpractical to test with the actual facilityload, permanent load banks should beconsidered in the initial power systemdesign, or a maintenance contractshould be considered with a serviceprofessional that can bring in aportable load bank to properly load thegenset during the exercise period.

At least once a year, all facilitiesshould exercise the power systemunder the actual facility load and full-emergency conditions to verify that thesystem will start, run and accept therated load. Running for up to severalhours under these conditions helps totest all the system components. (Itshould be noted that total operatingtime for testing may be limited by localauthorities for the purpose of reducingexhaust emissions released into theair.)

Besides verifying that the generatorset will start and run, periodicexercise has the benefit of heating updiesel fuel and eliminatingaccumulated condensation in the fueltank. Since clogged fuel filters andfuel contamination are among theleading causes of power systemmalfunctions, the cycling andrefreshing of fuel is an important stepin ensuring overall system reliability.

5. Code complianceThere are a number of industry andgovernmental codes that addressstandby generator set and powersystem reliability issues. Some affectthe manufacture of power systems,and some affect their installation,maintenance and operation.Compliance with all the appropriatecodes will increase reliability. Codesaddressing or impacting power systemreliability have been established by thefollowing organizations:

• NFPA (National Fire ProtectionAssociation) – Section 110addresses the standards forperformance for a standby powersystem and recommends monthlymaintenance and periodic testing.

• IEEE (Institute of Electrical andElectronics Engineers) – Definesreliability and addresses protocolsfor improving it through analysis andtesting.

• NEC (National Electrical Code) –Also known as NFPA 70, the NEChas become the de facto standardset of electrical requirementsthroughout North America. NECSection 700 sets standards forcommissioning of generator setsand sets operational parameters.

• JCAHO (Joint Commission onAccreditation of HealthcareOrganizations) – Recommendsminimum standards for standbypower systems for healthcareorganizations, including record-keeping, maintenance and periodictesting under load conditions toensure reliability.

• UL (Underwriters Laboratories,Inc.) – A national testing and ratingorganization. Compliance with theorganization’s UL 2200 code isdesigned to ensure that standbypower systems are safe. UL 1008 is arating for automatic transferswitches that verifies the switch willoperate reliably for at least 3,000operations – a number that is notlikely to be exceeded for many years.

• IBC (International Building Code) –Sets seismic standards forgenerator sets installed ingeographic areas prone toearthquakes to ensure reliableoperation after a seismic event.

• ISO (International Organization forStandardization) – Thisorganization’s ISO 9000 family ofstandards helps power systemmanufacturers develop qualitycontrol systems. ISO 8528 sets

standards for load acceptance andtransient response.

• EPA (U.S. Environmental ProtectionAgency) – Sets standards foremissions from many sources,including emissions from standbypower systems.

• Local air quality codes – Recent airquality laws enacted in the SouthCoast region of California arerestricting some generator sets torunning a maximum of 30 minutesper month. This practice may affectthe long-term reliability of standbypower systems by reducing thefrequency of power system testingand possibly damaging generatorsets by not allowing them to reachminimum operating temperature.Where local codes discourageproper generator set exercise due toair quality concerns, consult yourgenerator set’s manufacturer forrecommended exercise procedures.

ConclusionStandby generator sets are veryreliable machines with normalavailability in excess of 98 percent onan annual basis. However, thegenerator set is only one componentin a standby power system, andreliability needs to be considered interms of the total system design.

In addition, close coordinationbetween the facility manager and allthe power system equipment andbuilding automation system (BAS)suppliers during design, installationand commissioning is vital formaximizing reliability. Thiscoordination is necessary to identifypotential failure modes and developsolutions before problems occur. Byconsidering these factors along withthe generator set manufacturer’srecommendations, managers ofmission-critical facilities can beassured of the highest possiblereliability of their standby powersystems.


Michael Dauffenbach is Factory Training/Power Gen Specialist for MTU Onsite Energy. He can be reached at [email protected]


Leveraging Innovation toMeet Market DemandsInnovation has always been a drivingforce within the technology industry,and data centers are no exception.However, it is interesting howinnovative concepts have helped shapethe IT and data center landscape, andhow they continue to do so.

Many traditional data centers wereoften times created and constructedwith immediate need in mind.Companies needed to secure theirdata, and colocation facilities weredeveloped to help them address theirrequirements. While this worked verywell for the immediate needs of 10 to15 years ago, some things were notcarefully planned to be viable for thelong-term. Fast forward a decade orso, and we began to see a shortage ofavailable colocation space, along withother power constraints as technologycontinued to evolve and consumemore and more power. In today’s high-density colocation environment, it isnot unusual to have server racksconsuming 200+ watts per square foot.

This recent increase in new andinnovative supportive technologies andinfrastructure has many facilitiesscrambling to meet market demands.Instead of scurrying to simply meetclient requests, data centers caneffectively and efficiently leveragetheir design and infrastructure tosuccessfully support clients not onlytoday but for the long haul.

The question becomes, then, how hasinnovation allowed the industry tokeep up with demand, and whatinnovative methodologies andprocesses are now taking center stageto help ensure everyone’s needscontinue to be met?

SecurityHeightened security and compliance

measures utilized by data centersensure client’s vital information iskept private and safe at all times.However, how safe is safe and how doclients know that their data is trulyprotected? It’s one thing to havesecurity staff onsite, but thecompliance standards the industry hasseen develop assist in creating peace-of-mind for clients. Perhaps they arenot the first thing people think ofregarding data center or technologicalinnovation, but SAS 70, PCI DSS, andSSAE 16 in the near future, helpaddress many data center securitymarket demands. While the SAS 70standards have been around for quitesome time, their application to datacenter facilities is commonly seen as abenchmark of sorts for security. This,in turn, has created an environmentthat allows facility operators toaddress the demands of today’sincreasingly data-driven world, andcontinue to do so moving forward. TheSSAE 16 standards will effectivelyreplace SAS 70 in the next fewmonths, helping ensure U.S. serviceorganization reporting standardsmirror and comply with theinternational standard – ISAE 3402.

PCI DSS is a much more recentexample of an innovative conceptapplied to service providers, includingdata centers. Again, through theassessment of a facility’s securityfeatures by qualified auditors, a datacenter is able to present a suitablesolution to anyone that takes onlinepayments. Here we see how the rise ofe-commerce created a market need,and how a new concept, the PCI DSSstandards, helped address it. Theseare just two examples of keen securitycertifications that were developed andapplied to our industry. There will nodoubt continue to be others in the daysahead and the IT solutions provider or

data center that sets itself up toachieve them will continue to besuccessful.

InfrastructureInfrastructural support is a keycomponent of any data center and animportant aspect to potential clients.Additionally, this area has been andwill continue to be a driving forcewhen it comes to innovation. Thepower, cooling and network systems adata center leverages help to set itapart and distinguish it fromcompetitors. Not to mention, thesystems in place today can helpaddress the changing market needs oftomorrow.

A facility that is predicated upon state-of-the-art infrastructural designs willfind itself in an advantageous position.For instance, supportinginfrastructure that embraces modulardesign concepts enable a data centerfacility to not only address today’sneeds, but scale and expand asnecessary to meet futurerequirements. Leading infrastructureproviders are well aware of this andhave been developing systems andequipment in a manner to helpcapitalize on the ever-growing andever-changing market demands.Whether it is a modular chiller plant, amedium voltage UPS system, orsomething else entirely, innovationsmade at the manufacturer level shapehow a data center is able to bedesigned and successfully grow withevolving technology requirements.

Another area that infrastructuralinnovation has helped shape marketsis overall data center efficiency. Withgreen initiatives filtering into theindustry landscape, power usageeffectiveness (PUE) and energyefficiencies are increasingly critical to

26

By Ian McClarty

28

success. State-of-the-art power andcooling designs only get you so far.The supportive technology has to be inplace first, or the design will bemeaningless. Fortunately, there is noshortage of innovation within datacenter infrastructure systems.

CloudThere is maybe no technology concepttoday more talked about than cloudcomputing. With this increasinginterest, more and more businessesare looking into learning and investingin the unique product. Subsequently,many data centers are re-examiningcurrent technologies to prepare for thefuture and address market demands.However, instead of redesigning anentire system, which could requiresignificant additional capital

investments, data centers canleverage existing techniques, providedthey considered future needs duringthe design process.

Power and cooling needs for a cloudserver are going to be different thanthose of a colocation solution, and theunderlying support systems of a datacenter need to account for that. Casein point, a cloud server may only bepowered on for short periods of time,whereas a more traditional serverrequires a more continuous powersupply.

Although cloud may be a relativelynew topic, data centers can leveragecurrent infrastructure, design andsecurity arrangements to fit the needscloud computing requires, providedthe innovation within those systemsallows for it.

The Road AheadMeeting the current and ever-changing demands of clients and theirbusinesses is an ongoing process.However, with the right approach andmindset, they can be successfullyaddressed. Fresh innovations and newsystems are constantly beingdeveloped within the technology anddata center infrastructure space,helping ensure solutions meetrequirements. A few of theseinnovations have been highlightedhere, and who knows what the futurewill bring? One thing is certain, a datacenter that is open to or set up toaccommodate these innovations willcontinue to grow and adjust asnecessary to provide optimal ITsolutions.


Ian McClarty is the president of Phoenix NAP. He can be reached at [email protected]

30

Opening Comments

The industry is continuing to adaptto a changing landscape of efficiencymandates making energyconsumption one of the primaryconsiderations. Previously, reliabilityconsiderations dominated thedesign of mission critical spaceswith energy consumption being asecondary concern. The operationalenvironmental envelope within thedata center (temperatures andhumidity) can have a significantimpact on energy consumption.ASHRAE has continued to adaptalong with the industry byperiodically expanding their thermalenvironmental envelope guidelines.The following data points illustratehow periodic expansions ofrecommended temperatures haveevolved:

• ASHRAE 2004 – 68 °F to 77 °F

• ASHRAE 2008 – 64.4 °F to 80.6 °F

Beginning in 2004, ASHRAEpublished both Recommended andAllowable Ranges for IT equipmentthermal envelopes. The industry’slong term experience with reliabilitydominating decisions regardingcooling resulted in very fewinstallations and operatorsimplementing the Allowable Rangeeven though it created significantopportunities for both capital costand energy cost savings.

Another influence observed for notoperating within the AllowableRange was a lack of industryvalidated failure rate data whichcould identify the potentialconsequences of operating withinthis envelope. In May 2011, ASHRAE

published a Whitepaper thatprovides this failure rate data,information the industry has beenseeking for years. This Whitepaperprovides groundbreakinginformation that can radicallychange how we think about servercooling.

In order to responsibly pursue andapply these opportunities for CapExand OpEx savings, the followingbasics need to be understood:

• Where are the temperaturesmeasured?

• What are the differences in theData Center EnvironmentalEquipment Classes?

• What is an EnvironmentalEnvelope and why not a simplerange?


Pushingthe

Envelope -Expanding

EnvironmentalConditions In The

Data Center

By Don Beaty

• What are the definitions ofRecommended and AllowableEnvelopes?

In 2004, ASHRAE established thattemperatures for compliance withthe guidelines were to be measuredat the air inlet to the IT equipment.Figure 1 shows the measurementpoints at both the server and racklevel. Measurement at the serverlevel shows conformance with themanufacturer’s requirements andmeasurement at the rack or evenrow level is for systemic purposessuch as facility health.

Recommended Envelope:

The purpose of the RecommendedEnvelope is to provide guidance todata center operators onmaintaining high reliability whilesimultaneously operating their datacenters in the most energy efficientmanner. The RecommendedEnvelope is based on IT OEM’sexpert knowledge of server powerconsumption, reliability, andperformance vs. ambient temp.

Allowable Envelope:

The Allowable Envelope representparameters that IT manufacturersuse to test their equipment in orderto verify that functionality withinthose environmental boundaries.

There are many different uses andrequirements for both IT equipmentand data centers. To address thesedifferences, ASHRAE establishedEnvironmental Equipment Classes.In 2004, the categories were definedas Classes 1 to 4. In 2011, the datacenter Classes were modified to bedefined as Classes A1 to A4. TheseClasses vary the environmentalrequirements with A1 being themost stringent and A4 the least

stringent. Figure 2 provides anoverview of the amended 2011Classes as well as a comparisonwith the 2008 Classes.

Figure 3 provides the environmentalspecifications for each of theclasses including bothRecommended and Allowable. Thespecifications include parametersfor dry bulb temperature, % relativehumidity, dew point and rate of riselimits. There are variations as towhether % relative humidity, dewpoint, or both is listed. The intentwas to expand the environmentalenvelope and to most effectivelydefine a given environmentalcondition.

Explaining psychrometrics is notwithin the scope of this article butsome overall psychrometriccharacteristics are important tomention. Reliable IT equipmentperformance is dependent on bothtemperature and humidity and themost effective way to visualize thisrelationship is with a psychrometricchart (Figure 4).

The example chart includes dry bulbtemperature, wet bulb temperature,

32


Figure 1 – ASHRAE Thermal GuidelinesMeasurement and Monitoring Points

Figure 2 – ASHRAE Table 3: 2011 and 2008 Thermal Guideline Comparisons

34

dew point temperature, and %relative humidity conditions. Byplotting any two of these conditions,the remaining conditions aredetermined. There is anenvironmental envelope shown inFigure 4 for the RecommendedRange as well as for the AllowableRanges for all Data Center Classes.

The word “Allowable” in AllowableRange really says it all.Implementing the Allowable Rangepermits data centers to achievesignificant reductions in CapEx andOpEx based on the expanded rangeof environmental values. This wouldbe an easy decision if there was noconcern for the reduction of serverlife or server reliability by doing so.

Though it is anti-intuitive, there areactually situations where operatingin the Allowable Range does notreduce server life and serverreliability and may in fact, actuallyimprove these characteristics.

The fundamental basis for thisstatement is that server life andreliability are largely dependent ontheir individual operating conditions.While this may seem obvious, whatis not so obvious is how muchimpact the variation in conditionsand their cumulative impact canhave.

This situation is loosely analogousfor any equipment that may beexposed to an extreme condition fora short time period versus a longtime period. Ultimately, thecomparison of these operationalmodes can yield a significantdifference in life and reliability. Thisdescription can be expanded toencompass two conditions:

• Extreme optimum conditions

• Extreme harmful conditions

Consider the following threescenarios. The initial reaction mightbe that Scenario 1 is the best from a

life and reliability perspective.However, there is at least sometemptation to consider Scenario 3 asbeing the best from a life cycle andreliability perspective.

• Scenario 1 – 100% of lifetime atconstant “average conditions”

• Scenario 2 – 90% of lifetime at an“extreme harmful condition”

• Scenario 3 – 90% of lifetime at an“extreme optimum condition”

The ASHRAE Whitepaperdemonstrates, with somesupporting statistical data suppliedby manufacturers, that volumeserver hardware failure ratedecrease with a reduction intemperature. On the surface, thissupports the historical notion that“colder is better” with respect tooverall data center operatingtemperatures.

In general, colder being better is


Figure 3 – ASHRAE Table 4: 2011 ASHRAE Thermal Guidelines provides environmental specifications for equipment

36

true from a failure perspective aslong as a non-condensingenvironmental conditions aremaintained (above the dew point).However, using compressorizedcooling (traditional coolingequipment) to produce a colderenvironment may needlesslyconsume energy which cannegatively impact both CapEx andOpEx.

As an alternative to relying solely oncompressorized cooling, aneconomizer system may be used tosupplement and in some casesentirely replace the necessity formechanical cooling. A simplifieddefinition of an Economizer Systemsis one that takes advantage of theclimate outside being able to beused directly or indirectly providecooling without the use ofcompressorized equipment. TheASHRAE Allowable Range forclasses A1 to A4 enables an expand

number of hours per year that theclimate outside can meet theallowable conditions for proper ITequipment functionality.

Assume a data center is currentlyoperating with a constant server airinlet temperature of 70°F.Depressing the air inlet to lowertemperatures using compressorizedcooling may decrease thatequipment’s efficiency andsubsequently increase OpEx costs.However, if the outdoor temperatureis less than say 65 °F for a largenumber of hours per year, theopportunity exists to operate atserver air inlet temperatures below70 °F. This lower operatingtemperature provides better serverconditions from a life cycle andreliability perspective.

Figure 5 shows the temperatureimpact on the failure rates ofvolume server hardware. It is

important to note that these failurerates were assembled by ASHRAETC 9.9’s IT OEM subcommitteewhich included many IT OEMs. Thedata is derived directly from IT OEMsand provides valuable historicalperspective on real-world reliability.

Servers themselves are used formany purposes. These can includehighly critical applications to oneswith minimal criticality or even timesensitivity. The almost limitlessvariation in applications meansthere are different operationalstrategies amongst data centeroperations. This variation results inno single baseline applying to allapplications.

To resolve this variation in operatingprofiles, ASHRAE created a relativebaseline. That baseline assumes aconstant 68 °F operatingtemperature. This condition islabeled as an “X Factor” of 1. Wherethe temperature impact on volumeserver hardware failure rate is lowerthan the baseline, the “X Factor” isless than 1. Conversely, where thetemperature impact on volumeserver hardware failure rate isgreater than the baseline, the “XFactor” is more than 1.

It is important to note that the “XFactor” is relative. For example, if afacility operating at a constant 68 °Fhas 5 server failures / year, if the “XFactor” was 0.8, then the expectedfailure rate would be 0.8 x 5 serverfailures = 4 server failures / year.Conversely, if a facility operating at aconstant 68 °F has 5 server failures/ year, if the “X Factor” was 1.4, thenthe expected failure rate would be1.4 x 5 server failures = 7 serverfailures / year.

Significant opportunities exist formaintaining a server failure rateequal to or less than that createdwhen using traditional


Figure 4 – ASHRAE psychrometric chart shows how the environmental classes have expanded

38

compressorized environment isillustrated in Figure 6. The tableshows a time at temperatureweighted failure rate calculation forIT equipment, using Chicago bindata. This calculation shows thateven with temperature excursionsinto the Allowable range duringsummer extremes in Chicago, theaggregate effect of using acompressorless cooling system inthis location yields less failures.

This is an eye opening development.Returning again to the previousthree Scenarios, the Chicagoexample is actually a demonstrationof Scenario 3.

• Scenario 3 – lifetime where 90% ofthe time is “extreme optimumcondition”

In conclusion, the data centerindustry is currently in a highlyadaptive mode with manyfundamental precepts of design andoperation experiencing significantchanges. How we think aboutproviding cooling and subsequentlycooling system reliability needs tobe reevaluated in light of evolvingindustry data and mandates forgreater efficiency.. Embracing amuch broader view of potentialsolutions may result in significantCapEx and OpEx savings with nocompromise in reliability andperhaps potentially yielding a netincrease.


Don Beaty is President of DLB Associates. He can be reached at [email protected]

Figure 5 – ASHRAE Table C-1 Temperature impact on volume server hardware failure rate

Figure 6 – ASHRAE Table 7 Time at temperature weighted failure rate calculation for ITequipment in Chicago

40

Why DR?It’s about BusinessProcess AssuranceDisaster Recovery investment isessentially an insurance policy thatcritical business processes continuesat an acceptable service level after avariety of disaster scenarios. If thebusiness processes do not work underthe designed-for disaster scenarios,the DR investment is wasted and theenterprise is put at risk.

So a good DR plan starts with anexplanation of the business processesit is intended to protect and underwhat scenarios. This also helps yourbusiness leaders know what is atstake as you present DR costs andbenefits.

Business Process isSoftwareIn today’s workplace, most criticalprocesses of any scale are automated

by software. Whether you have 5employees and use QuickBooks or50,000 and use SAP it’s likely yourbusiness cannot operate properlywithout the invoicing, accountsreceivable, payroll or payablessupported by that software. Themanual processes such as enteringpaper invoices or printing paperchecks need to continue as well. Butwithout the software and the data,there’s little or no ability to do so. Nolonger is there a file cabinet with thepaper records to fall back on.

Knowing the business processes atstake allows you to identify whichsoftware applications need to beprotected, and to what degree.

Software needs HardwareAll software applications run on acollection of power strips, networkcables, switches, routers, servers,storage devices, load balancers andsecurity appliances. Today’s stack iscomplicated and can involve manydevices and configurations all of whichmust be in synch to operate properly.

To properly design a stack requiresunderstanding all its layers, from thepower strip at the bottom to theapplication at the top.

Since you know what software toprotect (based on the processes you’reprotecting) you can identify whichhardware to protect and you’ll knowexactly why you’re protecting thathardware. The cost of redundanthardware and data replication oftenexceed the perceived value, measuredby dollars or processes protected, of awarm DR site.

The threats to your hardware, andhence the scenarios under which youwant DR, can be described as aspectrum of damage from minimal toincapacitated.

FailureSpectrumWhat’s a Disaster?Instead of defining an exhaustive list of


CLOUD COMPUTING ANDDISASTER RECOVERY

By Yan Ness

42

hypothetical disaster scenarios at therisk of missing one, we use the failurespectrum to identify levels of failure.The failure levels are cause-agnosticwithin a stack, but they help us identifyhow we are (or are not) coveredagainst that degree of failure.

At the lowest level of the failurespectrum, a failed power supply in aserver or failed disk drive in an arraymight occur. Any application runningon a single server with a single powersupply and no backup is certainlyexposed to a catastrophic level offailure result in the case of a powersupply failure. A more severe degreeof failure would include a catastrophicfailure to an entire server (CPU, RAM,melt down etc).

At the highest level of the failurespectrum are things such as abuilding fire that destroys the stackand all its surroundings. Anyproduction environment withoutproper offsite backup couldexperience a complete loss of criticalprocesses and data records. A welldesigned DR plan must protect thecompany against failures at this end ofthe spectrum.

Failures at the lower end of thespectrum are more common thanthose at the top, but can be just ascatastrophic. So it makes sense toinvest in the lower ends of thespectrum first. And this is where cloudcomputing really helps DR.

Cloud and DRAll Clouds are Not equalWell-designed clouds should isolatethe virtual server from the hardware.This means that if the host hardwarehas a catastrophic failure, the virtualserver should still be able to run,albeit with a small delay. This is whatit means to isolate the virtual serverfrom the hardware host. But mostcloud offerings don’t isolate yourvirtual server from the host it runs on.In many cloud computing offerings, asingle host failure causes all virtual

servers on that host to fail. In otherwords, there is no hardware isolation.To use Cloud Computing to improveyour chances of surviving a disaster,you must have hardware isolation.

Hardware isolation is achieved bybuilding a cloud that has N+1 (or 2N)hosts and uses virtualizationtechnology that will automatically failover a server from one host to another.

DR from the CloudsOnce your environment is in a privatecloud, the Disaster Recoveryinfrastructure becomes much lessexpensive and less complex.

First, you are well protected against acatastrophic server failure, assumingyou have hardware isolation in an N+1server environment. Since this is morecommon than more serious damageshigher up the failure spectrum, thisalone can drastically decrease acompany’s risk of IT failure. In fact, a2010 study by the Aberdeen Groupshowed that cloud users experiencedmore improvement in decreasingdowntime events over the past year,decreasing 9 percent, while non-cloudusers decreased only 4.7 percent.Moreover, the impact of downtime wasalso decreased. The average length ofdowntime per disaster recovery eventwas 8 hours for non-cloud users, and2.1 hours by cloud users (nearly fourtimes faster).

But the bigger benefits of cloudcomputing come at the point ofexecuting the disaster recoveryinfrastructure. Just compare the costand complexity of a DR plan for 25

physical servers with 8 networkswitches versus a private cloud with 3hosts, 2 network switches and 1 SAN(RAID and redundant controllers).

Moving the 25 physical servers to anN+1 private cloud first protects againstmore common failures such ashardware or server failure. Butvirtualizing your servers alsoessentially converts your serverhardware to software. Each serverbecomes represented by a file that youcan copy and move around. So to DR aserver, you merely have to copy a file.It’s a lot easier – and cheaper - toreplicate a file than hardware. Betteryet, it makes your server hardwareagnostic, so you no longer have tomaintain exact duplicate hardwarestacks and all of the consequentsoftware patches. You have the optionof using a minimal resource footprintfor your DR stack, knowing that if ithas to become production, dropping inan extra server or storage willseamlessly increase its capacity.

A private cloud with a few hosts and aSAN can use tools like SAN to SANreplication, Site Recovery Manager byVMWare, VEEAM by VEEAMSoft, or anynumber of available products toreplicate the servers as files. Ideally,the virtual servers will go to a 2ndoffsite location in a way such that theycan be started quickly in the event of adisaster.

The morale of the story is, wait untilyou’ve moved your environment to aprivate cloud before implementing a2nd offsite Disaster Recoveryinfrastructure. You’ll find it much


44

simpler and cost effective.

We eat our own dog foodWe used to have a suite of about 20physical servers at each data centerfor our own corporate infrastructure.We have databases, monitoring,management, web, content andnumerous other servers that manageand deliver our services. It’s criticalthat the business processes andapplication those servers support areoperating normally at all times.

We used to use a complicated, offsitebackup process for the 20 physicalservers that would move the data toanother site. We had spare serversand a long, cumbersome, step-by-stepprocess for restoring data andapplications in the case of a disaster.As we added more physical servers(we got up to 28), we had to expandthat pool of available hardware at thesame rate – and maintain all of it.

We migrated 27 of the 28 physicalservers to a Private Cloud with 3 hostsand a SAN. It has capacity for aboutanother 20 servers. We put a secondPrivate Cloud at another data center.We use VEEAM to backup each host tothe second private cloud at the otherdata center. We have a 2 hour RPO and2 hour RTP for all 27 servers. We testit a few times a year, and the failoveris reliable and fast. Sometimes we useit to fail over just a few servers thatmake up a single application. Everytime we add a new virtual server toour private cloud, it automaticallycomes with DR coverage. Now ourincremental cost to have warm DR forevery new server is slight. Ourprotection against disasters issignificantly improved just by ourmove to the cloud.

SummaryWe’ve seen companies spend asignificant amount of money for a DRsite with redundant hardware and data

replication to protect againstproduction hardware failures. This is avery expensive way to protect againsthardware failure. Save everyone timeand money by first migrating to an N+1(or 2N) cloud. Then, at minimum,always invest in offsite backup. Thatway, when you get asked “When willwe be back up?”, you can at leastanswer “The good news is ‘not never’.The bad news is, I’m not sure when.”The most cost-effective thing you cando to increase the protection of yourbusiness processes is to move theapplications that support them to anN+1 (or 2N) private cloud. Thisprovides significant protection againstwhole server failures without the costof moving production from one datacenter to another and greatlysimplifies the upgrade in resiliencyfrom “Not Never” to “within 4 hours.”

When the time comes to implement awarm DR site, you’ll find that startingfrom the N+1 private cloud gives yourcompany a huge head start.


Yan Ness is Chief Executive Officer of ONLINE TECH. He can be reached at [email protected]

46

The American Society of Heating,Refrigerating and Air-ConditioningEngineers’ (ASHRAE) goal for the 2010edition was to reduce energy costs by30% compared to the 2004 version.Capital costs and return on investment(ROI) were not considered during theprocess.

Previous editions of this standard havebeen interpreted to exclude datacenter equipment from the energyefficiency requirements since theywere considered to be process loads.The new 2010 version specificallyincludes data centers and introducesadditional specific requirements ontotal system efficiencies and individualequipment efficiencies. Many featuresroutinely incorporated into otherproject types are now required to beprovided in data centers.

These new requirements will improvedata center efficiencies and result inoperational savings. But they will alsomost likely increase the initial cost ofthe systems. System designs willchange as a result of these

requirements but those changes willneed to be carefully designed so as tonot reduce the reliability andavailability of the data centeroperations.

ANSI/ASHRAE/IES Standard 90.1-2010is a consensus document. Itrepresents a best practices approachto energy efficient systems. It is not acode or a specific requirement until itis adopted as such by the AuthorityHaving Jurisdiction (AHJ). This can beat the local state or federal level. Thisis commonly done through theadoption of energy codes such as theInternational Energy ConservationCode (IECC). The present IECC-2009 isbased on the ASHRAE 90.1-2007standard. The IECC-2012 code ispresently being revised and willincorporate many provisions of theASHRAE 90.1-2010 standard. As thisenergy code is adopted by variousAHJs, the requirements for datacenters will become enforceable.

Other standards that may eventuallyadopt the new version are certification

programs for energy efficiency andsustainability such as LEED. Thecurrent draft for the updated LEED2012 includes a mandatoryrequirement to comply with ASHRAE90.1-2010.

There are multiple sections of thestandard that may impact the datacenter. This article will review these ingeneral and more specifically some ofthe electrical and mechanical systemsmost affected.

Architectural – BuildingEnvelopeThere have been changes andadditions to this section that willminimally affect the data center. Theenvelope requirements always appliedto the buildings where data centerswere located. One of the changesincludes a mandatory requirement fora continuous air barrier for allbuildings. This requirement isbeneficial to the operation of the datacenter. Other changes will affect


IMPACTS on Data Centers from thenew ASHRAE 90.1-2010 Energy Standard

by Dennis Julian, PE, ATD, DCEPWayne Drooks, PE

49


support spaces such as offices but willnot be of consequence to the datacenter itself.

LightingThe allowable lighting watts persquare foot have been reduced forinterior and exterior spaces and thecontrol requirements have beenexpanded.

Exterior spaces have been broken outinto five zones. The zones are based onthe areas surrounding the data centerand vary from rural undeveloped areasto major metropolitan areas. Theexterior lighting power allowances varybased on the zone where the facility islocated.

This could be an area of concern forproviding adequate security lighting inrural and less developed areas.Lighting fixture types will need to beselected based on lumens per wattand may increase the use of advancedlamps and technology including LEDfixtures.

Interior lighting power levels havebeen reduced from 10-18% dependingon the classification of the data centerareas. Designs will need to ensurethat adequate lighting levels aremaintained with these reduced powerlevels.

Transformers includingPDUsMinimum energy efficiencyrequirements for transformers 600Vand below have been added to thestandard. Transformers need tocomply with the provisions of theEnergy Policy Act of 2005. Efficienciesare to be measured per the testingrequirements of NEMA TP-1 2002.Essentially the testing emphasizesefficiencies at part load(approximately) 35% since that is atypical loading for general usetransformers. This also tends to betrue in data centers with redundanttransformers (PDUs). Efficiencies asmeasured by NEMA TP-1-2002 aregenerally required to be over 98% asindicated in the standard.

EPACT 2005 federal law is notidentical to the NEMA Standard TP-1.A major difference is that the federalenergy efficiency mandate does notexclude K-factor rated transformersor harmonic mitigating transformers,which NEMA TP-1 does exclude fromits scope. The federal law also doesnot specifically exclude retrofit orreplacement transformers from itsscope, as NEMA TP-1 does.

NEMA TP-1 certified transformers aretypically more expensive than non-certified units.

Heating, Ventilating andAir Conditioning (HVAC)SystemsMost equipment efficiencies have beenincreased from the previous standardand economizers are required in moreclimates. Many of these requirementsapplied to other project types but arenow specifically required in datacenters.

PipingVariable flow piping systems. Thestandard requires piping systems to bevariable flow. Constant flow pipingsystems are not allowed. Variablepumping systems controlled bydifferential pressure and exceeding 10hp, shall have the sensor located atthe most remote heat exchanger or atthe location requiring the greatestdifferential pressure. The maximumpressure set point shall not exceed110% of the required differentialpressure to achieve design flow.Where DDC controls are used the

differential set point shall be resetdownward based on valve positionsuntil one valve is nearly wide open.Exceptions are provided for systemsthat include no more than three valvesand where the minimum flow is lessthan that required by equipmentmanufacturers, such as chillers,where total pump system power is 75hp or less.

This requirement will require themonitoring of all the chilled watervalves by the DDC system and acontrol sequence to calculate therequired differential pressure in thesystem based on the position of all thechilled water valves.

Piping shall be insulated per Table6.8.3B except where the designoperating temperatures rangebetween 60F and 105F or where heatgain or loss will not increase energyusage.

It is common to not insulate dry coolerpiping which could range from 40F to110F. This may lead to increased costsfor insulation.

The standard has developed aminimum pipe size requirement. It is ageneral requirement based onstandard weight steel piping. Theminimum pipe sizes do not apply topiping that is not in the critical circuitat design conditions for more than30% of the operating hours.

Therefore, piping that experienceshigher flows during maintenance orabnormal conditions is not required tobe sized based on the standard.

kVA Efficiency kVA Efficiency

15 97.0 225 98.530 97.5 300 98.645 97.7 500 98.775 98.0 750 98.8112.5 98.2 1000 98.9150 98.3

Source: Table 4-2 of National Electrical Manufacturers Association (NEMA®) Standard TP-1-2002, ‘Guide for Determining the Energy Efficiency for Distribution Transformers’.

THREE-PHASE TRANSFORMER EFFICIENCY

50

The table includes the piping sizingrequirement from the standard andsome typical industry values that havebeen developed based on comparisonof initial cost, pumping energy, noisegeneration and system erosion due tofluid velocity. As can be seen from thechart there is not a lot of difference inthe smaller sizes but as pipeincreases to 6 inch or larger thedifferences can be substantial.

The pipe sizing in the standard shouldlead to a reduction in pipe frictionlosses and therefore reduced pressurerequirements (head) for the pumpthereby reducing operating costs. Thelarger pipe sizes will result inincreased costs for piping, valves,supports and installation.

Maximum flow rates are based onvariable flow piping systems operatingover 4400 hours per year.

ControlsSystem controls shall not permitreheat or simultaneous heating andcooling for humidity control. This is acommon occurrence in computerroom system designs that will nolonger be permitted.

If the new ASHRAE TC9.9 temperatureand humidity requirements arefollowed this should have minor effecton computer room design andoperation. It could be a concern on

tightly controlled humidity levels andin rooms with high humidity and lowinternal heat loads leading toovercooling.

Additional more sophisticated controlsmay be required to prevent thiscondition.

EquipmentOne of the mandatory provisions is

minimum equipment efficiencies. Inaddition to the previous tables forequipment efficiencies, Table 6.8.1Khas been added that lists minimumefficiencies for Air Conditioners andCondensing Units Serving ComputerRooms.

Equipment efficiencies must beverified through a certificationprogram or the equipmentmanufacturer shall install apermanent label stating that theequipment complies with therequirements of Standard 90.1.

CRAH units cannot be single speedconstant volume. Per the standard, airhandling and fan coil units with chilledwater cooling coils and supply fans 5hp or greater shall have two speed orvariable speed motors. As of January2012 this requirement also includesunits with direct expansion coolingand a cooling capacity of 110,000Btu/h (9.1 tons).

EconomizersEach cooling system that has a fanshall include either an air or waterside economizer. Climate zones arethe same as the IECC 2006 standard.Exceptions are listed for the following:

a. Exception a: Individual fan-coolingunits for the following (from Table6.5.1B):

1. No economizer requirement fordata centers in climate zones 1a,1b, 2a, 3a, 4a (essentially southeastUS and mid-Atlantic areas)

2. Cooling systems under 135,000Btu/h (11.25 tons) in zones 2b, 5a,6a, 7 and 8 (essentially theNortheast, eastern side of theMidwest and Alaska and somesouthern areas of Texas and NewMexico)

3. Cooling systems under 65,000Btu/h (5.4 tons) in zones(essentially the West coast and thewestern side of the Mid-west).

b. Exception c: Spaces humidified tosatisfy process needs, but it isstated that this exception does notinclude computer rooms.

c. Exception j: Systems primarilyserving computer rooms where:

1. The total design cooling load ofall computer rooms in the buildingis less than 250 tons and thebuilding is not served by acentralized chilled water plant, or

2. The room total design coolingload is less than 50 tons and thebuilding is served by a centralizedchilled water plant, or

3. The local water authority doesnot allow cooling towers, or

4. Less than 50 tons of computerroom capacity is being added to anexisting building.

d. Exception k: Dedicated systems forcomputer rooms where a minimumof 75% of the design load serves:

1. Those spaces classified as anessential facility.


2 1/2 68 4.56 40-65 2.7-4.4 30-50 2.0-3.43 110 4.77 65-115 2.8-5.0 50-90 2.2-3.94 210 5.29 115-240 2.9-6.1 90-190 2.3-4.85 250 4.01 240-440 3.9-7.1 190-340 3.1-5.56 440 4.89 440-700 4.9-7.8 340-550 3.8-6.18 700 4.49 700-1450 4.5-9.3 550-1100 3.5-7.110 1000 4.07 1450-2400 5.9-9.8 1100-2000 4.5-8.112 1500 4.26 2400-3500 6.9-10.0 2000-3200 5.7-9.2Pipes over 12 5.0 8.3-12.8 7.3-11.2inches

ASHRAE 90.1-2010 Clear Water Glycol and Water MixNominalPipe Size(inches)

VariableFlowSystem(gpm)

Maximumvelocity(fps)

Industry Design Flow for water(gpm)

Typical Industry Designvelocity forwater(fps)

Typical Industry DesignFlow forglycol(gpm)

Typical Industry Designvelocity forglycol(fps)

52

2. Those spaces having amechanical cooling design of Tier IVas defined by ANSI/TIA-942.

3. Those spaces classified underNFPA 70 (NEC) Article 708 – CriticalOperations Power Systems (COPS).

4. Those spaces where coreclearing and settlement servicesare performed such that theirfailure to settle pending financialtransactions could presentsystemic risk as described in “TheInteragency Paper on SoundPractices to Strengthen theResilience of the US FinancialSystem, April 7, 2003”.

Using exception k to eliminate theneed for economizers could impose ahost of additional requirements formechanical and electrical systems tocomply with the definition of Tier IV,essential facility or a COPS facilityincluding structural and architecturalconsiderations for robustness,survivability after an event andcompartmentalization.

Water economizers: system shall becapable of cooling supply air byindirect evaporation and providing upto 100% of the expected systemcooling load at outdoor airtemperatures of 50F dry bulb / 45Fwet bulb and below. Exceptions:

1. Evaporative water economizersfor systems primarily servingcomputer rooms: 40F dry bulb / 35Fwet bulb.

2. Dry cooler economizers forsystems primarily servingcomputer rooms: 35F dry bulb.

3. Systems where dehumidificationrequirements cannot be met usingoutdoor air temperatures of 50F drybulb / 45F wet bulb and where100% of the expected systemcooling at 45F dry bulb / 40F wetbulb is met with evaporative watereconomizers.

Pressure drops for the economizersystem are restricted to reduceadditional load on the pumps during

non-economizer system operation.

Integrated economizer controls arerequired to allow partial cooling evenwhen additional cooling is required tomeet the load.

This requires series installedeconomizers in lieu of the morecommon parallel economizers.

Many water cooled chiller systemshave been installed with plate andframe heat exchangers that aredesigned and operated as a chillerreplacement when the outdoorambient conditions are appropriate.This implies the heat exchanger isinstalled in parallel with the chillersand the water is typically bypassedaround the chiller. Therefore, the heatexchanger is an all or nothingoperation.

This standard will require the heatexchanger to be in series with thechiller at all times so it can providepartial cooling.


CLIMATE ZONES

54

HumidificationSystems with hydronic cooling andhumidification systems designed tomaintain a minimum dew point of 35Fshall use a water side economizer ifan economizer is required.

Fan System PowerLimitationFan cooling systems exceeding 5 hpmust be variable speed and the motorsize must not exceed the next largerstandard size greater than the brakehorsepower (bhp). The bhp must beidentified on the design documents forverification by the AHJ.

System CommissioningThe standard requires that HVACcontrol systems be tested to ensurethat controls are calibrated, adjustedand are working properly. For projectslarger than 50,000 sf of conditionedspace, detailed instructions forcommissioning the HVAC systemsshall be provided by the designer inthe plans and specifications.

Energy Cost BudgetAlternativeSection 11 of the standard outlines therequirements to evaluate complianceof the proposed design with theprescriptive requirements of the

standard. There are still somemandatory requirements that need tobe met. Once these are satisfied,alternative designs can be evaluatedto show compliance with the standard.

ConclusionFollowing ASHRAE 90.1-2010 willprovide additional energy savingsduring operation. It may also lead tohigher initial costs for equipmentinstallation and structure. Designerswill need to be creative inincorporating these energy savingfeatures while keeping initial costsunder control and maintaining therequired level of reliability andavailability of the data center.


Dennis Julian, PE, ATD, DCEP, is Principal of Integrated Design Group. He can be reached at [email protected] andWayne Drooks, PE, Senior Mechanical is the engineer at Integrated Design Group. He can be reached at [email protected]

56


Data CenterInfrastructureManagement (DCIM): It’s one of the hottest buzzwords fordata center operators today. In 2010, areport by the Gartner Group predicteda 60 percent market penetration ofDCIM by 2014. In the report, “DCIM:Going Beyond IT,” analyst David J.Cappuccio explores the importance ofDCIM for dealing with rising energycosts and inefficient power usage inthe data center. The report indicatesdata center managers cannot reducecosts unless they have real-time datacenter software to provide an accurateview of current resources consumed.

Simply put: DCIM software is apowerful monitoring solutioneffectively integrating informationfrom a broad set of tools that havetraditionally been held within IT orFacilities management “silos.” Inaddition, DCIM solutions areinstrumental in reducing the time-to-deploy new servers by up to 50percent, extending the life of a datacenter by up to five years, as well ashelping to attain a power usage

effectiveness of 2.0 or less. And to capit off, with regards to extractinginformation from often disparatesystems, DCIM gives the data centerteam a “single lens” through whichthey can monitor capacity planning,risk management, power utilizationand overall efficiency of the datacenter.

A number of established data centersoftware players have alignedthemselves with this designation.However, not all DCIM solutions arecreated equal. Data center operatorslooking to deploy DCIM must conductproper due diligence before investing –it’s a key step to ensuring the solutionfits specific needs, now and in thefuture.

Investing in DCIMSoftware - Best Practices Before investing, the first step is todefine the challenge. Managing datacenter environmental performance isnot just about wires, black boxdevices, monitoring hardware, or thevast array of specialty devices being

pitched as cure-alls. It’s aboutmanaging all of the data such as:

• IT assets and system

• Space utilization

• MEP system resiliency andperformance

• Energy, financial and environmental

Unless there’s access to all theaforementioned data in a holisticformat – the data center’sperformance picture is incomplete.Why? Because the data is oftenstranded across multiple proprietarysystems. Identifying what data isavailable and where the gaps are isneeded to establish an optimalperformance management process.

Ideally, the best DCIM tool shouldprovide centralized monitoring,management, and intelligent capacityplanning of a data center’s criticalsystems. Furthermore, the platformmust conduct real-time monitoringand management across IT andFacility infrastructures to maximizedata center ROI.

Basic features and requirements DCIM

Promises, Promises: Does Your DCIM Solution Deliver?

by Fred Dirla

58

tools must have are:

• Identifying Primary User(s) –determine the number of users thetool needs to serve. Is it in thehundreds, thousands or millions?Make sure the tool can handle theload.

• Establish a Clear Outcome – identify focus areas i.e.environmental performance, energymanagement, efficient IT equipmentdeployment, capacity management.

• Power Cooling Information – aconstantly moving target, datacollection and analysis of powercooling information can posesignificant challenges. Take a smalloperation of about 1000 servers, witheach server having two plugs tomonitor. Add in rack-leveltemperature monitoring plus theability to monitor core power andcooling infrastructures – the wholeset-up could easily reach 3000measurement points. With pointschanging constantly, it might benecessary to measure and store avalue for each data point at afrequency of once per minute. That’sover 1.5 billion measurements peryear!

• Information Accessibility – a majorhurdle for most solution providers.DCIM must function as a truebrowser-based application thatallows full access to anyone on thenetwork. While many claim to offertrue DCIM, they either force users toinstall client software on eachdesktop or make them log into theapplication through a browser, whichmeans limited functionality.

• Scalability – the application mustallow room for growth as theorganization expands. The idealsolution should make adding newdata center facilities to the system asimple matter of adding to theexisting application. This will providevisibility into all data centers in theportfolio in a clear and consistent

manner regardless of scale ordesign.

However, to go beyond just the basics,add these DCIM elements to yourchecklist:

• Fully Browser-Based and Secure –This offers a powerful, real-timedata collection and database enginethat allows data center teams tomanage IT assets while supportingthe infrastructure to maximizeresource utilization and energyefficiency. Most importantly, the toolmust meet the most stringent datasecurity, application resiliency, andbandwidth requirements ofcorporate IT networks, while beingvendor neutral – allowing seamlessintegration with virtually anyoperations systems and hardwarerunning in the data center.

• Able to Handle Unlimited Users –Scalability is good, but unlimitedusers are better. The interface mustprovide rapid and intuitive access toall data as needed – from managinghundreds to hundreds of thousandsof circuits. It should also provideinformation down to the individualcircuit to prevent overload, whiletracking energy usage by the rack,row, or cabinet – providing simplePUE calculations for specificsystems.

• Holistic Data Views – A centralized"Single Pane View" throughoutvarious monitoring and controlsystems, installed at a single facilityor throughout an entire corporateportfolio is a “must have.”

• Real-Time Reporting Tools – Thesetools are needed for delivering alertsprior to a catastrophic event as wellas enabling pro-active planning forgrowth and expansion with up-to-the-minute information. In addition,the tools should provide a LoadSimulator to analyze load impactbefore cabinet equipment is added.From global reports, right down toinformation on individual circuits,

information must be delivered ineasy-to-read graphical depictions.

• Fully Compliant – DCIM solutionsneed all proposed levels ofmonitoring according to DOElegislation. A real-time datacollection engine helps data centeroperators manage IT assets and thesupporting infrastructure, tomaximize resource utilization andensure energy efficiency. An intuitivefront-end needs to provideimmediate access to vast amounts ofreal-time data for informeddecision-making.

• Easily Deployed – A flexible DCIMtool enables new assets to be addedwith little or no external programrequirement. This includes: powerutilization efficiency such as PUE,DCiE, RCI and facility load; profile-based analysis, defining a definitivestart point to understand a “greenstrategy” endpoint; and energysavings tracking with documentedresults.

• Out-Of-The-Box Functionality –Finally, a good DCIM tool offersuser-friendly capabilities to createcustom screens and reports. Theselected DCIM software mustleverage business intelligence toensure raw data is transformed intoactionable intelligence. This shouldinclude the ability to graph data andtrend it over time.

As energy prices continue to skyrocketand data centers become increasinglycomplex, managers are scrambling tofind new ways to bring costs andefficiencies under control. DCIMpromises to be the panacea for all thatails the data center. But beware: Notall DCIM tools are created equal. It’simportant to define objectives ahead oftime and ensure the tool offers thebest of the best in DCIM, including: aunified view, real-time reporting, fullcompliance, and ease of deployment.Only by doing your homework, can yourealize the true benefits of DCIM.


Fred Dirla is CEO of FieldView Solutions. He can be reached at [email protected]

60

Fire protection installations in DataCenters/Telecommunications roomsare being challenged by computingdensity increases and the push forPUE improvements. High cooling loadsand reducing cooling energy cost isdriving data center managers tochange the physical structure of theirrooms. Hot Aisle and Cold Aisle(HACA) containment systems in thevarious flavors they come in are veryefficient at improving coolingefficiency and can often have shortpayback timeframes. Many datacenters have had systems retrofittedas well as systems designed for newinstallations. It is obvious that they arehere to stay and for good reason.

HACA containment systems affect firesuppression and detection in manyways. They can make fires moredifficult to detect and the obstructionsthey create can make them moredifficult to control or extinguish. If youhave employed HACA strategies inyour data center, but have not madeadjustments for the fire protection ofthe room, you most likely have aproblem and it may be bigger than youthink.

This article attempts to describe howHACA systems will affect conventionalapproaches to fire protection in thedata center/telecom environment andconsiderations you should make whendesigning a room with HACA systems.The truth is the fire protectioncommunity has not kept up with thepace of change in technology rooms.Often code development followsbehind changes in technology.However, the good news is thatexperts within the fire protectionindustry are taking aim at thischallenge today in order to developguidance for the changing data center.

THE MISSIONCRITICAL FIRE RISK

The perception of risk to your facilitycan often be diminished by false

reasoning in two ways. First is thenotion that fires in data centers don’thappen. Companies are often silentabout fires in their IT facilities, inhopes of maintaining a positive brandimage. In addition many fires that dooccur in these buildings aresuccessfully controlled orextinguished by fixed fire suppressionsystems and are thus hardlynewsworthy. These facts make it verydifficult to monitor fire activity in datacenters; yet we know they occur. Fireprotection service firms respondseveral times a year to re-armsuppression systems that dischargedbecause of a fire. In the interest ofclient confidentiality these serviceproviders do not share informationabout fires.

The second deception in judging firerisk often occurs to data centermanagers who are investigating HACAsystems to employ in their facility.Manufacturers of containmentsystems are aware that the fireprotection community has concernswith the effects on fire detection andsuppression, so the containmentsystem literature often makesassurances that their system will workwith existing fire protection systems.Some of these claims are misleadingand are made from a simpleunderstanding of how fire suppressionsystems work, rather than beingbased upon solid fire science,experience and testing.

CONTAINMENTSYSTEMS AS NEW

BARRIERS IN THE DATACENTER

When you start to break down theconcepts employed by the varioustypes of containment systems they areessentially barriers or new partitionswithin the space. These barriers serveto direct the cooling airflow to where itis needed most at the face of thecomputer server.


By Lee Kaiser, PE Hot Aisle/Cold Aisle

Systems Driving Changes

to Fire Protection

62

Many people assume that if a barrieradded as part of a containment systemis transparent, it will not affect the fireprotection in the room. However, thesebarriers create three conditions thatchange the approach needed for fireprotection in a medium to high densitydata center:

1. The airflow pattern changes anddisrupts the normal development ofthe smoke plume.

2. The airflow velocity in the roomincreases within the contained aisleand can be a challenge forconventional means of smokedetection.

3. The barriers act as obstructions tosprinkler spray patterns and cleanagent suppression nozzles.

When thought of as barriers, the needfor a professional analysis of theroom’s fire detection and suppressionsystem becomes well warranted.

CHALLENGES FORDETECTION

The National Fire Alarm and SignalingCode (NFPA 72) gives guidance toengineers on the spacing of smoke

detectors in rooms with different airchange rates. The fire alarm code onlyprovides data for smoke detectorspacing in rooms up to 60 air changesper hour (ACH); this equates to datacenters loaded to roughly 5 kW perrack. According to the IntelCorporation, HACA containmentsystems start being used in datacenters with densities of near 12 kWper rack and higher. With coolingairflows sufficient for typical HACAcooling loads, air change rates withinthe contained aisles range from 500 to1000 ACH and higher. These high ratesmean high velocities and willchallenge ceiling mounted spot smokedetectors due to the velocity of the airand dilution of smoke.

It has been common for fire protectionengineers to specify air samplingsmoke detection (ASSD) in roomsexceeding 60 ACH because of theirincreased sensitivity to smoke.Because of higher velocitiesexperienced within contained aisles itmakes even more sense to utilizeASSD. Engineers should considerinstalling ASSD sample pipes/portsarranged to sample the hotreturn/exhaust openings in thecontained aisle. This would be donethe same manner as is commonlyapplied to traditional CRAC unit returnair grilles.

Data center designers would do well

to take notes on best practices ofsemiconductor manufacturing cleanrooms where ASSD is often employed.Clean rooms have similar challengesof high velocities, turbulent air flows,and directional routing of air. There isan easy translation to data centerswhere air sampling detectors shouldbe installed at the return air inlet toair handling units and somewhereimmediately downstream of the hotside of the server at the ceiling level.

Despite the fact that more research isneeded on how to detect fires in highairflow environments, manyprofessionals believe that thedetection techniques needed arealready available to the industry.

CHALLENGES FORSUPPRESSION

Whether the barriers that form theHACA containment system are appliedhorizontally, vertically or both, theycan affect sprinkler patterndevelopment and clean agentdispersion. NFPA 13: Standard for theInstallation of Sprinkler Systems isvery explicit in how to apply firesprinklers to overcome obstructions inthe protected space. These rulesshould be applied to sprinklers whereHACA barriers are applied. Many ofthese containment systems haveprovisions for “automatic” removalwhen a fire occurs; usually by meansof a fusible link. For removal to workthe fire has to grow to a point where itcan melt the link(s). If the link thatremoves the barrier is not placedperfectly over where the fire starts,the fire will need to grow larger tobuild heat in the location of the link.This also applies to systems thatrequire fusing of multiple links forbarrier removal. Be wary ofcontainment systems that require alarge fire to remove the barrier beforethe fire sprinkler system is given theopportunity to activate.

Plastic drop out panels, used mostoften in cold aisle containment


64

systems, are another type of barrierthat “automatically” removes. Thesepanels are UL listed and melt around135°F so that the sprinklers above thepanels can operate. Data centermanagers should know that the listingrequires use of standard responsesprinklers above the panel thatoperate at a higher temperature near155°F. Typically this sprinkler type isnot installed in data centers; usuallyquick response sprinklers which fuseat 135°F are used. The temperaturedifference is important and could leadto issues with sprinklers operatingbefore the necessary panels havedropped out. Installing quick responsesprinklers inside the contained aislemay be a good way to avoid this issue.

In facilities utilizing clean agent fireextinguishing systems, containmentbarriers must be removed prior toagent release. Containment systemswhich rely upon fusible action forremoval are a problem because of thelarge fire size needed to obtain the

action. Clean agent systems in datacenters most often activate upondetection of smoke, not heat; and aredesigned to extinguish smalldeveloping fires. The problem ofbarrier removal can be overcome byadding extra clean agent nozzleswithin the contained aisle.

Clean agent nozzles have several ofthe same obstruction distancerequirements as sprinklers. In HACAretrofits a qualified fire protection firmshould be consulted to ensure therequired extinguishing concentrationcan be obtained given the new barriersinstalled in the space.

It is a valid assumption that cleanagents will disperse to spaces whichare not in line-of-sight of the agentnozzle, such as the ability to reach theinside of server cabinets. The highairflows associated with containmentsystems challenges our currentassumption and more research mustbe done on this topic.

NEW FIREPROTECTION

RESEARCH INTO HACAENVIRONMENTS

The movement to use HACA in missioncritical environments has not goneunnoticed by the fire protectioncommunity. Recently there has been asignificant emphasis on this issue bythe National Fire ProtectionAssociation (NFPA) and their researchdivision, The Fire Protection Research

Foundation (FPRF). In March 2011, theFPRF held their Suppression,Detection and Signaling TechnicalWorking Conference and took a day forindustry experts to focus on the issueat a meeting “Fire ProtectionChallenges in Telecommunicationsand Information Centers.” In April theNFPA 75 and 76 TechnicalCommittees, which address IT andtelecommunications rooms, held ajoint meeting to continue theconversation from March. Newresearch papers on the topic of fireprotection in data centers werepresented in June at the 2011 NFPAconference.

Most recently a new research projectis being considered by the FPRF tostudy smoke detection design in highairflow environments. This is certainlyexciting for the fire protectioncommunity and will hopefully provideclarity to the issues surrounding HACAcontainment systems.

While more research is needed intothese new challenges from allinterested groups, one thing is forcertain; hot aisle/cold aislecontainment systems have theattention of data center managers,designers, and fire protectionprofessionals. All are working toensure a reliable means exists todetect and suppress fires in theseenvironments. If you have added HACAto your data center and not had aqualified professional evaluate yourfire protection system, you could berisking higher losses than what yourbusiness can tolerate.


Lee Kaiser, PE is the Engineering Manager for Orr Protection Systems. He can be reached at [email protected]

66

The International Building Code (IBC)is a comprehensive set of buildingstandards that was first proposed in1997 by the International Code Council(ICC) and adopted in 2000. The IBCsought to harmonize the manynational, state and local codes thatgovern the design of structures in aneffort to eliminate duplicative orconflicting standards and, therefore,make compliance more uniform.

The IBC has been updated on a three-

year cycle; the latest version is IBC-2009. Currently, all 50 states and theDistrict of Columbia have adoptedversion IBC-2000, IBC-2003, IBC-2006or IBC-2009 as their de facto buildingcode.

While the main focus of the IBC isstructural integrity and fire prevention,certain provisions govern thecertification and installation ofemergency standby power systemsused in locations that are seismically

active or are subject to high windloading of up to 150 mph. Dependingon the classification of the structureand type of occupancy, seismicallycertified emergency standby powersystems are required in order toensure power after a catastrophicevent, such as an earthquake or windevent.

The primary needs for electricalpower after such an event is for thecontinuing operation of essential


HOW THE IBC SEISMIC AND WIND-LOADINGSTANDARDS APPLY TO EMERGENCYSTANDBY POWER SYSTEMS

Understanding provisionsin the latest edition of theInternational BuildingCode is critical forspecifying emergencystandby power systemsthat will continue tooperate after events such as an earthquake or a hurricane.

By Dwight Wells

68

facilities to support the communityand various life-safety systems thatsupport building egress. Where powersystems are required for continuedoperation of the facility, the standbygenerator set and supportingcomponents must be sized to operateall other critical components in thebuilding such as: air handling units,air conditioning, cabinet heaters, airdistribution boxes, boilers, furnaces,chillers, cooling towers, water heatersand other similar mechanical,electrical or plumbing equipmentrequired to keep the buildingfunctional. Where life-safety is ofconcern, the emergency standbypower would be required to operateemergency lighting, elevators,

ventilating systems, communicationsystems, alarms, fire pumps and othersystems involved in protecting life-safety. At a minimum, IBC certificationand installation details are required inseismically active locations for thefollowing essential facilities that areclassified as Occupancy Category IV inTable 1:

• Hospitals with surgical oremergency treatment facilities

• Fire, rescue, ambulance and policestations

• Designated public storm shelters

• Emergency response centers

• Power-generating stations and

public utilities

• Structures with toxic or hazardoussubstances

• Aviation control towers and air trafficcontrol centers

• Facilities involved in critical nationaldefense functions

• Water storage facilities required forfire suppression

Additionally, an emergency powersystem that continues to operatefollowing a seismic event plays apositive role in business continuity,allowing the proper shutdown ofmanufacturing processes or thepreservation of computer data – bothof which help reduce financial risk.

Deciding when to specifya seismic power system Not every area of the U.S. or type ofstructure is required to have aseismically certified emergency powersystem. According to the IBC, aseismically certified emergency powersystem is only required in locationsand structures that meet certaincriteria. Figure 1 shows the areas inthe country that are seismically activeand where seismic design must beconsidered. The criteria includeimportance factor (Ip), buildingoccupancy category, site soil class andspectral response acceleration.

• Importance Factor – The IBC uses animportance factor (Ip) to designatewhether an emergency standbypower system is a critical or non-critical component. A non-criticalcomponent has an Ip of 1.0, but acritical component has an Ip of 1.5when any of the following conditionsapply:

1. The emergency standby powersystem is required to operate afteran earthquake for life-safetypurposes (such as egress lighting,sprinkler systems, fire protectionsystems, smoke evacuation, etc.).

2. The structure contains hazardousmaterials.


Table 1

70

3. The emergency standby powersystem is located in an OccupancyCategory IV structure and is neededfor the continued operation of thefacility or its failure could impairthe continued operation of thefacility.

• Occupancy Category – Table 1 showsthe occupancy categories ofbuildings and other structures aslisted in the IBC-2003, 2006 and2009. Categories I through III do notrequire a seismically certifiedemergency power system unlessthey are located in a seismicallyactive area with short-periodresponse acceleration greater than0.33g and the equipment is given anIp = 1.5 because of number points 1or 2 above. See Table 2. However, allCategory IV structures require sucha system when the importancefactor is 1.5 (i.e., essential) and SDSis more than 0.167g.

• Site Classification – In anyseismically active zone, the potentialfor structural damage is influencedby the soil type. The least structuraldamage can be expected on solidrock (Site Class A), while the moststructural damage can be expectedon loose, liquefiable soils (Site ClassF). See Table 3.

• Short-Period Response Acceleration– This is a number (SDS) derivedfrom the expected ground movementforces (measured in g = accelerationdue to gravity) in seismically activelocations as defined by the UnitedStates Geological Survey (USGS).The value also accounts for the soiltype of the location. Refer to Figure 1

as a reference map of the contiguousUnited States. The higher the SDSvalue the more severe are theseismic forces acting upon astructure and its contents. Thisnumber is then used in conjunctionwith an Occupancy Category (I – IV)to determine a Seismic DesignCategory (A through F). Buildingswith Seismic Design Categories C –

F have requirements for seismicallyqualified components when thecomponent Ip = 1.5.

The following three criticalparameters are the basis fordetermining whether a seismicallycertified emergency power system isrequired:

• An SDS of 0.167g or greater

• Occupancy Category IV with an Ip =1.5

• Seismic design category of C, D, E orF and a component Ip = 1.5

Power system structuremust also resist windloadingThe IBC also addresses wind loadingand its effect on the performance of anemergency standby power system. Forthose states that have adopted theIBC-2003, IBC-2006 or IBC-2009, thebuilding (or enclosure) that houses theemergency standby power systemmust resist any overturning forcescaused by expected high winds, andthe generator set must stay mountedto its foundation and be operable afterthe event in all Occupancy Category IVstructures (essential facilities).


Figure 1. Map is based on SDS values assuming Soil Site Class D. Areas is green, brown and bluerepresent areas of the United States affected by the seismic requirements of the IBC Codes.(Map compliments of The VMC Group)

Table 2

Table 3

72

Figure 2 shows the areas of thecountry where high wind loadingneeds to be considered in the designof structures that house emergencystandby power systems. The minimumwind speed for design of structures inthe United States is 85 mph.

Seismic designresponsibilityAccording to the provisions in IBCstandards, the entire design team isresponsible for making sure anemergency standby power systemstays online and functional after aseismic or high wind loading event.This group includes: emergencystandby power system manufacturers,suppliers, installers, design teammanagers, architects and consultingengineers. Each has a critical role toplay in making sure that structuraland nonstructural componentsperform as designed.

IBC Chapter 17, the ContractorResponsibility states:

Each contractor [i.e., all members ofthe design team listed above]responsible for the construction of a

main wind- or seismic-force-resistingsystem, designated seismic system ora wind- or seismic-resistingcomponent…shall submit a writtenstatement of responsibility to thebuilding official and the owner prior tothe commencement of work on thesystem or component. Thecontractor’s statement ofresponsibility shall containacknowledgment of awareness of thespecial requirements contained in thestatement of special inspection.

Seismically certifiedemergency powersystemsIt falls to the emergency standbypower system manufacturer to providea product that is certified to withstandthe typically expected seismic andhigh wind loading forces and tocontinue operating after a seismicevent has occurred. The provision inIBC-2009, Section 1708.4 SeismicCertification of NonstructuralComponents states:

The registered design professional[usually the architect, consultingengineer or electrical contractor] shall

state the applicable seismiccertification requirements fornonstructural components anddesignated seismic systems on theconstruction documents.

1The manufacturer of eachdesignated seismic systemcomponent subject to the

provisions of ASCE 7 Section 13.2.2shall test or analyze the componentand its mounting system or anchorageand submit a certificate of compliancefor review and acceptance by theregistered design professionalresponsible for the design of thedesignated seismic system and forapproval by the building official.Certification shall be based on anactual test on a shake table, by three-dimensional shock tests, by ananalytical method using dynamiccharacteristics and forces, by the useof experience data (i.e., historical datademonstrating acceptable seismicperformance) or by more rigorousanalysis providing for equivalentsafety.

2Manufacturer’s certification ofcompliance for the generaldesign requirements of ASCE 7

Section 13.2.1 shall be based on


Figure 2

74

analysis, testing or experience data.

An emergency standby power systemconsists of a base, engine, alternator,fuel supply, transfer switches,switchgear and controls. While theengine generator set is naturally arobust piece of equipment, designingfor survival of a seismic event alsofocuses attention on the generator setmounting to the foundation andexternal attachments, such as fuellines, exhaust and electricalconnections.

To certify the components of anemergency standby power system, thegenerator set and its associatedsystems are subjected to acombination of three-dimensionalshake-table testing and mathematicalmodeling. The IBC requires that these

tests be performed by an independent,approved, third-party supplier that canissue a seismic certificate ofcompliance when the seismicqualification is successfullycompleted. Once a particulargenerator set passes the seismicqualification, it is the responsibility ofthe manufacturer to label theequipment, indicating the seismicforces to which the equipment wassubjected. Figure 4 illustrates a typicallabel on a seismically certifiedgenerator set.

Local regulations may bemore stringent While the general provisions of the IBC

have been largely adopted as the defacto building code in many states andlocalities, the project engineers shouldconsult with local jurisdictions toverify that all applicable localstandards have been met. InCalifornia, for example, the Office ofStatewide Health Planning andDevelopment (OSHPD) has set seismicstandards for hospitals and healthcare facilities in accordance with boththe 2007 California Building Code andIBC. While these local codes andrecommended seismic testingprotocols are largely harmonized withIBC, consultation with local authoritiescan reduce the risk of installing asystem which may ultimately prove tobe non-compliant.

ConclusionsWhile the IBC addresses all facets ofstructure design and construction inall 50 U.S. states, it also addresses theperformance of a number ofnonstructural systems, such asemergency standby power systems.The IBC’s requirements for emergencystandby power systems are intendedto ensure that structures withincertain occupancy categories will haveemergency power after a catastrophic

event, such as an earthquake or windevent. As such, it has set seismicdesign and testing standards for themanufacturers of emergency standbypower systems.

All members of a structure’s designteam – emergency standby powersystem manufacturers, suppliers,installers, design team managers,architects and consulting engineersneed to be aware of the seismic andwind loading provisions within IBC foremergency standby power systems.Power system manufacturers haveundertaken advanced design andtesting programs to comply with theseismic provision within IBC involvingthree-dimensional shake-tabletesting, finite element analysis,mathematical modeling andexperience data. Certified powersystems are labeled as having passedseismic testing by a qualified,independent testing organization. Byworking with a power systemmanufacturer that offers seismicallycertified products, the structuraldesign team can be assured that it willhave an emergency standby powersystem that will perform as designedafter a seismic or high wind loadingevent.


Dwight Wells is Product Manager for MTU Onsite Energy. He can be reached by [email protected]

Figure 4

“a seismicallycertified emergencypower system isonly required inlocations andstructures thatmeet certaincriteria...”

77


INTERESTED IN PRESENTINGAT THE SPRING CONFERENCE?Visit www.7x24exchange.org and download the Call for Presentations

DEADLINE: JANUARY 27TH

SUBMIT AN ARTICLE FOR THE SPRING 7x24MAGAZINE

Visit www.7x24exchange.org and download the Call for Articles

DEADLINE: FEBRUARY 6TH

2011 FallEND-TO-END RELIABILITYMISSION CRITICAL FACILITIES:

SPRING 2012June 10 – 13, 2012Hilton Orlando Bonnet CreekOrlando, FL

FALL 2012November 11 – 14, 2012Arizona BiltmorePhoenix, AZ

Interested in presenting at 7x24 Exchange? Complete the Call for Presentations at www.7x24exchange.org or call 646-486-3818 x104

S A V E T H E D AT E ! ! !

For information about sponsoring a 7x24 Exchange eventplease contact Brandon Dolci, CMP at (646) 486-3818 x108

LEVERAGINGINNOVATION

2011 Fall ConFerenCe

HigHligHts

LEVERAGINGINNOVATION

The Fall Conference themed “End-to-End Reliability: Leveraging Innovation” will be held November 13-16 atthe Arizona Biltmore in Phoenix, AZ. The Fall Conference will feature compelling keynotes, a high levelpanel, concurrent sessions, a spectacular vendor event, and more...

General Hugh Shelton, Fourteenth Chairman of the Joint Chiefs of Staff, will kick off the conference with asession entitled “Leadership That Leaves a Legacy”.

The President of MTechnology, Steve Fairfax, will open the second day with a keynote presentation entitled“Hidden Threats to Data Center Security”.

Kevin Kealy, Security Architect for AT&T will provide the closing keynote “Control Systems – The New Target”.In keeping with the theme, additional presentations on leveraging innovation will be delivered with topicssuch as:

• AOL – Managing Disruption: A Data Center Love Story

• Deutsche Bank – ECO Data Center

• EPA – Enabling Energy Savings in Data Centers with ENERGY STAR

• Facebook – Open Compute Program

• Green Grid – Data Center Maturity Model

• Uptime Institute – Global Data Center Concerns Affecting the North American Market

• Panel – The Hunt for Talent

• Optimizing Cloud Uptime: Data Centers and the Hybrid Cloud

• Planning the Key to Replacing Critical Cooling in a Live Data Center

In addition to enhanced programming 7x24 Exchange International presents “An Evening At Corona Ranch”.Join us on as we venture off property to Arizona’s finest Mexican rodeo and fiesta destination. During afestive dinner reception, 7x24 Exchange guests will catch a glimpse of the sun setting on the vistas ofcolorful cacti as you sip your favorite cocktail, taste tequila, visit with a cigar aficionado while enjoying a handrolled cigar and participate in many activities that will include a variety of casino games, Mexican games andmusical entertainment. After dinner, it’s time for the Charreada Rodeo Show.

This event has been made possible thanks to the following partners:

ABB, Active Power, ASCO, Caterpillar, Clark Mission Critical, ComRent, Cummins Power Generation,

Cyberex, Data Aire Inc., DPR Construction, FieldView, GE, Gilbane, HP, IM Corp, Kling Stubbins, Kohler,

Layer Zero, Mitsubishi, Monico, MTU, Opengate, PageSoutherlandPage, PDI, Russelectric, S&C,

Schneider Electric, Siemens, Skanska Mission Critical, Staco Energy, Starline, Stulz, Syska Hennessy,

Virginia Economic Development Partnership, Walker Engineering, Whiting-Turner

For the complete Fall Conference program and registration information please visit 7x24exchange.org or call (646) 486-3818.

2011 SPRING CONFEREN

CE CORPORATE LEADER

SHIP PROGRAM PARTN

ERS

(AT

PR

ESS

TIM

E)

MEDIAPARTNERS

GOLDPARTNERS

PLATINUMPARTNERS

SILVERPARTNERS

BRONZEPARTNERS

MITSUBISHI ELECTRIC

UPS Division

BECOME INVOLVED IN YOUR LOCAL 7x24 CHAPTER

BCAB

SK

ND

ON

MT

WYSD

NE

CO

NM

KS

LA

OK

WI

IA

IL

TN

MS

TX

AR

MO

MN

ME

ID

WA

OR

CA

NV

AZ

UT

MI

NCKY

INOH

SCGA

FL

AL

WV

MARICTNJDEMDDC

NHVT

NY

PA

VA

BCAB

SK

ND

ON

MT

WYSD

NE

CO

NM

KS

LA

OK

WI

IA

IL

TN

MS

TX

AR

MO

MN

ME

ID

WA

OR

CA

NV

AZ

UT

MI

NCKY

INOH

SCGA

FL

AL

WV

MARICTNJDEMDDC

NHVT

NY

PA

VA

VISIT WWW.7x24EXCHANGE.ORG TODAY TO PARTICIPATE IN YOUR LOCAL CHAPTER

Attention end users and vendors...

INTERNATIONAL CHAPTERS INCLUDE:

Arizona

Atlanta

Canada

The Carolinas

Delaware Valley

Empire State

Florida

Greater Washington DC Area

Lake Michigan Region

Lone Star (Dallas)

Metro New York

Midwest (Kansas City, MO)

New England Area

Northern California

Northwest (Seattle, WA)

Rocky Mountain

Southeast Michigan

Southern California

Texas South

82

MEMBER ADVERTISING RATE CARD

7x24MAGAZINE OPPORTUNITIESFollowing are the Editorial Guidelines together with the Member Advertising Rate Card. Advertisersinterested in placing an ad may fax the insertion order to 7x24 Exchange at 212.645.1147 or emailto [email protected]. Questions? Please call Jeremy O’Rourke at 646.486.3818x109.

Advertiser indemnifies 7x24 Exchange against losses or liabilities arising from this advertising. 7x24 Exchange assumes no liability whatsoever,except to the extent of a one time paid advertisement of the same specification, in the next or similar publication, if any proven or admittederrors or omissions have occurred. Payment is due upon receipt of the invoice. Interest shall be charged at 2% per month compounded to yield26.82% per year on overdue accounts. Revisions to previously submitted ad copy are subject to additional charges. A charge of $30.00 will belevied for returned checks. In the event of a contract cancellation, the advertiser or agency agrees to repay 7x24 Exchange any discountsgranted for multiple insertions less any discount applicable for the number of insertions completed in the contract. All cancellations must bereceived in writing prior to the advertising sales deadline. All premium positions are non-cancelable. Prices are net of agency commission.

BLACK AND WHITE RATES

Size 1X 2X 3X

Full Page $1,500 $1,300 $1,1002/3 Page 1,100 1,000 9001/2 Page Island 900 800 7001/2 Page 700 600 5501/3 Page 600 550 5001/4 Page 500 450 400

COVERS & PREMIUM POSITIONS – INCLUDES 4 COLOR

Size 1X 2X 3X

DPS $5,000 $4,500 $4,0002nd / 3rd Cover 2,500 2,200 2,0004th Cover 3,500 2,750 2,500

Full

Page

1/4

Verti

cal

1/4

Hor

izon

tal

1/3

Verti

cal

1/3

Hor

izon

tal

1/2

Verti

cal

1/2

Hor

izon

tal

1/2

Islan

d

2/3

Verti

cal

2/3

Hor

izon

tal

Live Area: 7.5” x 10”Trim Size: 8.5” x 11”Bleed Size: 8.75” x 11.25”Halftone Screen: 133 lines up to 150 linesDPS Mechanical Requirements:Live Area: 16” x 10”Trim Size: 17” x 11”Bleed Size: 17.25” x 11.25”Halftone Screen: 133 lines up to 150 lines

8 1/2” x 11” MECHANICAL REQUIREMENTS

COLOR RATESProcess Color (4/c): $900

Revisions and Proofs: $50

Position Guarantee: 15% premium

*Non-Members add 40% to all rates

NON-BLEED AD DIMENSIONSSize Width LengthFull Page 7.5” 10”2/3 Horizontal 7.5” 6.5”2/3 Vertical 5” 10”1/2 Island 4.875” 7.5”1/2 Horizontal 7.5” 4.875”1/2 Vertical 3.625” 10”1/3 Horizontal 7.5” 3.25”1/3 Vertical 2.5” 10”1/4 Horizontal 4.5” 3.25”1/4 Vertical 3.25” 4.5”

EDITORIAL GUIDELINESManuscript specifications: Feature articles vary in length from 500 to 2,000words. While we accept articles in a variety of formats, it prefers to receivematerials on CD. All articles must be received by the deadline to beconsidered for a specific issue. Material submitted after the deadline will beconsidered for the following issue.

Bylines: All articles should include a brief (1-2 sentence) author biographicalsketch at the end of the article, that includes the author’s name, title,affiliation, address, and phone number. Photos of authors are never used. Wedo not pay authors for contributions.

Visuals: Authors are encouraged to submit photographs and charts, graphs,or other illustration that will help readers understand the process beingdescribed, though it does not guarantee that visuals will be used with thearticle. Submit all charts, graphs, and other artwork separately; do notincorporate them in the body of the article. Indicate caption materialseparately. We reserve the right to publish submitted visuals.

Editorial proceduresAll articles are reviewed for suitability. Accepted materials are then edited forgrammar and to conform with our editorial style. All attempts are made topreserve the author’s writing style, however, we have the right to edit for style,clarity, and to fit space allotments, and to make final selection on headlines,subheads, and graphic treatment. Manuscript submission implies authoragreement with 7x24 Exchange’s Editorial Policies.

CopyrightWe require first serial rights for submitted articles. This means the author(s)grant us the right to publish the article for the first time. We also requestpermission for electronic distribution on 7x24 Exchange’s web site,www.7x24exchange.org.

DisclaimerThe responsibility for accuracy remains with the author. The opinions andinformation in bylined articles in this publication are those of the authors anddo not necessarily reflect those of the Officers and Board of Directors of 7x24Exchange.

Press optimized PDF or EPS accepted.No JPEG or TIFF accepted.

Convert all spot colors to CMYK.

Do not use 4 color black. All 4 colorblack will be replaced with 100%black at the advertiser’s expense.

Fonts must be embedded or outlined.

Maximum ink density 280%

••••••••••••••••••••••••savethedateJune 10 – 13, 2012 I Hilton Orlando Bonnet Creek I Orlando, FL

MEMBER ADVERTISING RATE CARD

Promises, Promises: Does Your DCIM Solution …...floor model takes years. In addition, if a company...

Documents

Transcript of Promises, Promises: Does Your DCIM Solution …...floor model takes years. In addition, if a company...