tipson creatinga sustainable datacenter · the time to ﬁ nd out more by contacting your pdi sales...

THE MAGAZ INE OF 7x24 EXCHANGE INTERNAT IONAL

S P R I N G 0 8

TIPS ONCREATING ASUSTAINABLEDATA CENTER

TTaacckklliinnggTTooddaayy’’ss DDaattaaCCeenntteerr EEnneerrggyyEEffffiicciieennccyyCChhaalllleennggeess

Newslink_S08_11fx.q7:Newslink_S08.q5 4/11/08 2:31 PM

Power Distribution, Inc. | 4200 Oakleys Court | Richmond, VA 23223800.225.4838 | 804.737.1703 fax | web site: www.pdicorp.com

©PDI PDU 2/08

PDI’s WaveStar™ monitoring system will help you pro actively manage your data center loads. By helping you manage your power distribution equipment better with the WaveStar™ monitors and PDI’s patented BCMS (Branch Circuit Monitoring Systems) systems, your PDIq (Power Distribution Intelligent Quotient) is higher! With PDI’s management system you get real time data on every PDU, RPP and breaker in your entire data center – whether it’s a PDI unit or someone else’s.

Data is gathered and sent to a local on-fl oor monitor or to a central location. PDI’s proven BCMS system makes all this possible, so now is the time to fi nd out more by contacting your PDI

sales representative.

So, if you need the best in WaveStar™ Static Switches, PDUs or RPPs and the ability to resolve load and system

confi guration problems before they happen, the solutions are there

to raise your PDIq.

Put PDI’s Smart Monitoring Systems to work for you today and increase your .

pdu pdu

WaveStar ™ switch

rpp rpp

rpp

rpp

central computer

jcomm tm jcomm tm

q

q

Newslink ad 08.indd 1 2/27/08 1:14:00 PM


3SPRING 2008

Contents6 Tips on Creating a

Sustainable Data Center

10 Tackling Today’s Data Center Energy Efficiency Challenges

16 15 Seconds Versus 15 Minutes: Designing for High Availability

18 The Role of Data Center Professionals in Business Continuity Planning

22 RIGHT-SIZING: Current Criteria for Data Center Multi-MarketSite Selection

26 Parallel Optics Changes the Direction of Network Cabling

32 Inside 7x24

DIRECTORS & OFFICERS

Chairman of the Board Robert J. CassilianoBusiness Information Services, Inc.

President/Chapter RepresentativeWilliam LeedeckeVanguard

Vice PresidentRoy L. ChapmanAmerican Express

DirectorDavid SchirmacherGoldman Sachs & Co.

Vendor RepresentativeRuss B. MykytynCampbell Company

Administrative DirectorKathleen A. Dolci(646) 486-3818 x103

Membership & EducationTara Oehlmann, Ed.M.(646) 486-3818 x104

ConferencesBrandon A. Dolci, CMP(646) 486-3818 x108

www.7x24exchange.org

322 Eighth Avenue, Suite 501New York, NY 10001(646) 486-3818

Power Distribution, Inc. | 4200 Oakleys Court | Richmond, VA 23223800.225.4838 | 804.737.1703 fax | web site: www.pdicorp.com

©PDI PDU 2/08

PDI’s WaveStar™ monitoring system will help you pro actively manage your data center loads. By helping you manage your power distribution equipment better with the WaveStar™ monitors and PDI’s patented BCMS (Branch Circuit Monitoring Systems) systems, your PDIq (Power Distribution Intelligent Quotient) is higher! With PDI’s management system you get real time data on every PDU, RPP and breaker in your entire data center – whether it’s a PDI unit or someone else’s.

Data is gathered and sent to a local on-fl oor monitor or to a central location. PDI’s proven BCMS system makes all this possible, so now is the time to fi nd out more by contacting your PDI

sales representative.

So, if you need the best in WaveStar™ Static Switches, PDUs or RPPs and the ability to resolve load and system

confi guration problems before they happen, the solutions are there

to raise your PDIq.

Put PDI’s Smart Monitoring Systems to work for you today and increase your .

Newslink ad 08.indd 1 2/27/08 1:14:00 PM


4 SPRING 2008

Chairman’s Letter

As Spring approaches it is the perfect time for us to be thinking Green!

Over the years there have been a number of “Hot Topics” of interest that havespanned numerous 7x24 Exchange conferences. Some of you mayremember when we were focused on harmonics and the problems causedby non-linear loads, or power density and our insatiable thirst for talkingabout watts per square-foot. How many debates have we had on thesetopics?

These days a major topic of interest is energy efficiency or “Greening” of datacenters. The rapid rise of energy consumption relating to data centeroperations has the attention of everyone from data center operators togovernmental regulators. Availability of power, environmental impact relatingto increasing energy consumption, and rapidly increasing operationalexpense are all key factors in the debate.

On September 18, 2007 Dave Schirmacher, 7x24 Exchange Director, and Iattended a session on energy efficiency at the New York Stock Exchangehosted by the New York State Energy Research and Development Authority(NYSERDA). The purpose of the meeting was to spread awareness, solicithelp from major corporations, and to emphasize the increasing role thatgovernment agencies like the US Department of Energy (US DOE) and theEPA will play in driving solutions.

The 7x24 Exchange is committed to leading the industry with regards toeducation and information sharing on the important topics of the day. To thatend our 2007 Fall conference presented 5 topics on energy efficiency andour upcoming Spring 2008 conference scheduled for June 1 – 4 at the BocaRaton Resort and Club in Boca Raton, Florida will have 12 presentations onGreening and will feature a panel of experts on Data Center Efficiency.

As an organization the 7x24 Exchange provides value to you and yourcompanies by ensuring focus on the most important challenges you face.

I look forward to seeing you at our Spring Conference!


6 SPRING 2008

David F. Anderson, PE, PMP

TIPS ON CREATING ASUSTAINABLE DATA CENTER

Everywhere I go, from the grocery store to the car pool parkingplace outside IBM’s mega data centers, I am now reminded to begreen. Whether it be purchasing alternative “green” products for

everyday use or taking care of business in a sustainable way, we nowhave the opportunity to change for the better. I simply define better asdoing operations with less energy and creating longer lasting value.Green is all about doing the right thing and saving costs. How canyou spruce up your Data Center? Here are some tips:

1. Begin with an Enterprise goal in mind. Create lasting greenness.Small scale will yield small results. Large scale will yield bigger results.Often every project looks good and green by itself, but when added upa sub optimal data center and complex infrastructure has been created.IBM has a vision that can become a green blueprint for a data centerthat has state of the art capabilities while using less energy and space.The New Enterprise Data Center exploits virtualization, servicemanagement with automation and aligns with business goals. Thejourney of transformation includes being simplified, shared anddynamic. Start a green program with goals in mind.

2. Exploit virtualization to reduce the number of servers and improveflexibility. Virtualization or using software / hypervisor technology torepresent virtual servers rather than physical servers is a very greentechnology. Reducing the number of power drawing components in thedata center to a minimum directly slashes the amount of energyconsumed as well as reduces the cooling requirements. LogicalPartitioning (LPAR) and virtualization technologies such as VMware,PowerVM and z/VM to name a few break the physical boundaries ofservers and drive up utilization reducing the need for many servers.Starting with pilot or proof of concept projects is easier than everbefore since IT vendor services and virtualization technologies on allplatforms have matured. From open source virtualization to mainframez/VM virtualization, implementation services and technologies aboundto start eliminating the wasteful approaches of one server perapplication and a variety of servers for every production server. Many

applications can be hosted on a physical server and still have theautonomy because of virtualization. How many servers can beeliminated? Compression rations of 1 to 8 are common and 1 to 50compression ratios are often achieved with the best virtualizationtechnologies.

3. Exploit virtualization to reduce the amount of storage networkingequipment. This includes SANs and using virtual I/O connectionswithin servers such as IBM’s HiperSockets, Virtual Ethernet for Systemsp and i, and OSA integrated layer 2 and 3 switching.

4. Use integrated approach to server consolidation to optimizesavings. More than a single methodology needs to be applied to getthe fewest number of servers. IBM’s Enterprise Computing Model forreducing thousands to about 30 Large Centralized servers used thefollowing approach to consolidation:

� Migrate servers delivering largest savings first (i.e., strandedinfrastructure). This primes the pump and generates enthusiasm andsavings for other green projects.

� Eliminate assets with lowest utilization first. These assets are notpulling their weight when measured by watts/logical image or othercommon metrics to compare servers.

� Identify assets with an upcoming compelling event to mitigateexpense (upgrade, move, asset refresh). It is always easier to havea positive ROI and be green within the normal refresh of assets.

� Aggregate by customer work portfolio to leverage strong customerbuy-in. Ease of migration assists speed and successful workloadmigrations.

� Start with oldest technology first as it uses the most power andprovides the least performance.

� Focus on freeing up contiguous raised floor space. This enablesgrowth and the addition of energy efficient new IT and facilitiesequipment.

� Provision new applications to the mainframe or another largecentralized server.


5. Drive to high utilization rates.Virtualization and management ofworkloads is key. The OperatingSystem must manage W/L to businesspriorities and dispatch in an automatedmanner. The average Wintel server isused only 5-15% of the time. Nomanager would allow his people towork 5-15% of the time. With newtechnologies and automation utilizationrates can go beyond 50% and at thesame time improve flexibility andresponsiveness as more resources canbe tapped for peaks.

6. Consolidate on large servers. Fewerlarger servers will convert AC to DCmore efficiently than many smallerservers with smaller and less efficientpower systems. Power supplies oflarge servers are capable of operatingat very high efficiencies (+90%). Theability to more efficiently shareresources makes running on a fewlarger Systems more efficient thanmany small ones.

7. Eliminate redundancy but keep highavailability and disaster recoverycapabilities. High availability anddisaster recovery can efficiently and ina green way be designed into serverconfigurations. Engines can now addnon disruptively to almost all platformsreducing the need for extra servers. Nolonger is an idle server needed for“what if” scenarios. Production serverscan back up other production servers.Configuring the ability to non-disruptively add (and reduce) capacityfor production or disaster recoverywithout having idle or underutilizedservers significantly reduces the size ofthe footprint and slashes the energyconsumed in the data center.Commonly used technologies includeIBM’s On/Off Capacity on Demand (addengines by the day) and CapacityBackup Upgrade (CBU) for DisasterRecovery. A data center can be greenedand the bottom line affected by usingfewer servers while having the ability toincrease capacity without adding serverand the associated facilityinfrastructure.

8. Measure and put the costs ofenergy where they are incurred.Automated measuring and billing ofenergy consumption makes usage partof cost and green decisions. Withoutenergy and cooling knowledgerequirements are unknown, inaccurateand often overplanned, leading toinefficiencies. An example of newtechnology to optimize energy use isIBM’s Active Energy Manager (AEM).Monitoring energy usage and

The following are some additional green ideas for sprucing up your data center.Some you may have already done. Others may yield small to massive energy savings.All of them I have observed in various data centers in the last 12 months.

• Hot and cold aisles with 2 floor tiles width. Stop mixing hot and cold air where possible.• Use free cooling. Outside air can substantially reduce energy required by Computer Room AirConditioners. Data center site selection will enable more days of free cooling when climate has bigdifference between day and night temp. Colorado is an example of an excellent climate to exploit freecooling.

• Enable Active Energy Management (IBM Power Director Active Energy Management with Tivoli). If you do not measure you miss understanding easy opportunities to improve.

• Larger than code copper distribution (wiring for data center).• Enable dynamic provisioning of server and storage resources. This can reduce as well as turn offnumber of servers drawing power.

• Modeling suggests at least 24 inches of unobstructed raised floor. Optimize air flow putting lessstress on CRACs.

• Plan for water. It will be used on high end equipment to reduce energy requirements and hot spots onthe raised floor. Putting water closer to heat loads will minimize the need for more air conditioning.

• Control hot air rising. Ceiling return of hot air. Row air curtains. Block for recirculation (but make surecan use sprinklers). Dividers to the ceiling can prevent hot air from escaping into cold aisles.

• Plugged openings (cables, power) not in cold aisle. Block at server or other IT equipment cabletailgates and cable openings in the raised floor to prevent cold air losses and improve efficiencies.

• Two tiles on both hot and cold aisles. Enables both tiles to be pulled up to ease of access tounderneath floor. Having tiles with many perforations (holes) is a must.

• Tile and CRAC placement can be optimized (temperature and air flow) with fluid dynamic analysis.Moving air with less direction changes is best.

• Cables overhead – even power, Leave under floor to pumping and air flow. When laying out cablesmake sure they do not impede airflow if under raised floor (or above). A tray or trough can be made.

• Auto lights out. Lights need to be on only when someone is inside data center.• Shock and vibration support for racks. This is a must for earthquake regions and can be plannedwhen refreshing equipment or greening the data center.

• Lighting on back of racks for ease of servicing. • New generation battery, flywheel and diesel generator back up. Flywheel can provide a very greenway to keep up power system until generators start. Equipment has substantially improved in energyefficiency in last year. Use latest generation of UPS, flywheel and generators. Newer generations ofUPS are much more efficient than older generations.

• Redundancy design for power and cooling. Eliminate common cause failures where ever possibleincluding UPS by having flywheel and batteries rather than redundant battery UPS.

• Liquid cooling for hot equipment with rear door heat exchanger or side car technologies.• Enough water cooling taps built into water system for new growth.• Minimize impedance of piping (water or air) to reduce pumping power required.• Leak detection under raised floor for water distribution system.• Variable frequency drives on all pumps and Air Conditioning equipment.• Easily displayed power and thermal monitoring for data center. Large display to highlightsuccess/energy saved. Take pride in using less.

• Where possible, measure overall data center Power Unit Efficiency and plan on ways to keepreducing as new state of the art equipment and processes become available.

• Physical: Fire protection system FM – 200 or wet sprinkler + Security system with “man trap” to keeppotential intruders from entering the raised floor.

• Liquid-side Economizer. Efficiently control humidity.• Thermal storage to optimize use of chillers and reduce energy cost at peak hours. IBM’s “cool battery”cooling solution can improve energy efficiencies and reduce electric bill substantially. IBM Bromontsaw a 45% improvement using stored cooling technology.

• High efficiency pumps, chillers and fans for cooling towers. • If possible, use cooling towers in summer and reuse waste heat to reduce energy in winter.• Use Virtualization for testing and potentially actual Disaster Recovery. Where possible architect andbuild active active solutions (production in 2 places that can non disruptively add capacity to backup and scale when needed).

• Link facilities and IT. Offer new types of service level agreements based on power/performance notjust performance.

• Set both tactical and strategic green goals. Educate team and have the entire team devoted toachieving green goals.

• Celebrate your green successes.


1-800-225-5250 www.russelectric.com

At Russelectric, special requirements or unusual designs are not interruptions of our business... they are our business.

Russelectric specializes in using proven technologies and components in innovative ways to satisfy even the most exotic customer requirements. We’ll never compromise on quality, safety, or performance. And, above all, we’ll never ask our customers to settle for less than exactly what they want.

Independently owned and operated, Russelectric has only one goal… to take care of our customers.

POWER CONTROL PEOPLE YOU CAN RELY ON

“

Russelectric custom engineers every automatic transfer system to specific customer requirements

Our facilities staff wanted state-of-the-art automatic electronic control plus the ability to override it and operate the switches manually if the situation called for it.

Russelectric supplied both in all our switches.”

08-103 Resized AutoManual (for 71 1 2/14/08 2:22:37 PM

8 SPRING 2008

Tips on Creating a Sustainable Data Center

developing trends is key to understanding how energy is being used.This first step to optimizing energy use opens up the potential tobecome more efficient and optimizing for performance / watt. Managingenergy use is an evolving concept in the data center. Capping power atthe server level and optimizing to deliver the right performance perwatts can be achieved using AEM. In the future the most efficient datacenters will treat servers like you treat lights in your house, turningthem off when not in use or at least turning on only what you need.Linkage to Total Cost of Ownership (TCO) ensures green is part of everydecision. Benchmark the entire data center as well as local areas forcontinuous improvement. Use commonly accepted methodologies suchas the Power Usage Efficiency ratio (Total power / IT equipment power= PUE) from Green Grid Consortium or the Energy Efficiency ratio.

9. Use the concept of hierarchical storage. Picking the right mediumand format for storing data is like picking the right vehicle for a trip.Not every trip needs an 18 wheeler or a motorcycle. A combination ofdisk, tape and hybrid technologies optimizes the use of energy whilegiving your data a secure and extendable home. Tape, a greenstorage equipment star, uses the least amount of energy and shouldbe part of the storage constellation. Disk storage should be fordemanding applications that require frequent updates. The Virtual TapeServer can mask latency with many applications and is another greenstar in the storage constellation. Larger and slower disks use lessenergy and if their latency can be masked the energy efficienciesgained by their use is worth it.

10. Use the latest equipment. Newer generations of IT equipment aremore energy efficient and give better performance than older ITequipment. Begin greening your data center with replacing the oldest,most inefficient equipment first. Newer generations of servers andstorage are built with more efficient power supplies, processors,

memory and I/O. Just about everything in newer servers and storageprovides more performance or stores more data with fewer watts. Weall have experienced how digital cameras have provided morememory, functions and better performance in the last 5 years. Serversand storage are on similar technology improvement trajectories.Servers scale higher in performance while using fewer watts perlogical image. Decommissioning older servers that never weredesigned for virtualization or energy efficiency can be one of the mostcost effective ways to green your data center. Like the gas guzzlingclunker that needs to be replaced with a hybrid there are better waysnow to run applications. There is a BIG difference between IT and yourautomobile. The new servers allow you to replace 8 to 80 servers. Thesimplicity in running fewer physical servers is a no brainer. TheSystem z10, IBM’s new generation of energy efficient large centralizedservers, can replace approximately 1800 distributed servers. Theenergy efficiency and space savings enable the data center to addcapacity within the four walls.

11. Do not wait for the 11th hour. Start to a sustainable data centernow. The biggest savings of all for going green is to create a cultureand infrastructure that exploits technology for creating a sustainabledata center. With green concepts and projects a data center can growin capability / capacity while continuing to use the same or less spaceand energy. Conducting or having a 3rd party conduct an energy auditcan benchmark where you are and identify projects with ROI’s that canbe prioritized to give cascading green returns. For every watt you savein IT equipment you reduce the infrastructure (UPS, cooling, etc.) loadand generate savings for future projects.

Newslink MagazineHalf-Page Horizontal7-1/2 x 4-7/8, 4-Color

With Sentry! Smart CDU

Switched CDU

Server Technology, Inc.1040 Sandhill Drive

Reno, NV 89521USA www.servertech.com

[email protected]

toll free +1.800.835.1515tel +1.775.284.2000fax +1.775.284.2065

©Server Technology, Inc. Sentry is a trademark of Server Technology, Inc.

Solutions for the Data Center Equipment Cabinet

With Sentry!

How Do You DistributePower in Your DataCenter Cabinet?

Sentry Single or 3-Phase products with 100-120V, 208-240V or 230V AC voltages & 20A, 30A or 60A input currents. They provide power distribution solutions from high density applications to high density server clusters.

try!CDU Product Family: Basic, Metered, Smart & Switched

Metered CDU

Smart CDU

> Local Input Current MonitoringMetered CDUMetered CDU

> Reliable & Economical> Reliable & EconomicalBasic CDU

Smart CDU

Switched CDU

> Local Input Current Monitoring> Supports External Temp. & Humidity Probes> IP Monitoring of Power Temp. & Humidity

Switched CDU> Local Input Current Monitoring> Supports External Temp. & Humidity Probes> IP Monitoring of Power, Temp. & Humidity> Remote Power Control for Each Outlet:

ON / OFF / Reboot

David F. Anderson, PE, PMP is the Green Consultant of IBM Corporation. He can be reached at [email protected]


1-800-225-5250 www.russelectric.com

At Russelectric, special requirements or unusual designs are not interruptions of our business... they are our business.

Russelectric specializes in using proven technologies and components in innovative ways to satisfy even the most exotic customer requirements. We’ll never compromise on quality, safety, or performance. And, above all, we’ll never ask our customers to settle for less than exactly what they want.

Independently owned and operated, Russelectric has only one goal… to take care of our customers.

POWER CONTROL PEOPLE YOU CAN RELY ON

“

Russelectric custom engineers every automatic transfer system to specific customer requirements

Our facilities staff wanted state-of-the-art automatic electronic control plus the ability to override it and operate the switches manually if the situation called for it.

Russelectric supplied both in all our switches.”

08-103 Resized AutoManual (for 71 1 2/14/08 2:22:37 PM


Tackling Today’s Data Center Energy Efficiency ChallengesA Software-Oriented Approach

James Parker, PE, CEM, CMVP and Bill Brown, PE

10 SPRING 2008

INTRODUCTION

Driven by the sheer amount of consumed electrical powerand related year-over-year increases, growing scrutiny is beingplaced upon electrical efficiency in data center applications. Inthe year 2000, servers and their associated cooling equipment inthe United States (not counting network and storage equipment)consumed approximately 23 billion kWh.1 By 2005, this figurehad risen to 45 billion kWh, accounting for 1.2% of all 2005 U.Selectricity sales.1 The total magnitude of data center energyconsumption, including network and storage equipment, is nowestimated to be between 1.5% and 3% of all electricitygenerated.2 This increased energy consumption also comes withan increased price tag: Currently, the typical 3-year cost(operating expenses + amortized capital expenses) of poweringand cooling servers is approximately 1.5 times the cost of theserver hardware itself, with projections of up to 22 times by theyear 2012.2

The reasons for increased energy consumption in the data centerare, simply put, increased demands for computing power andnetwork storage.

REDUCING ENERGY COSTS THROUGH INCREASED ENERGY

EFFICIENCY

Data center energy consumption reduction may be carriedout in a number of ways. Reducing the IT electrical load is anobvious way of reducing energy consumption, since the IT loadalso affects the amount of cooling power required to remove heatfrom this equipment. However, manufacturers of semiconductorchips and the servers that use them are already increasing thecomputing throughput per watt such that it roughly doubles everytwo years.3 The increasing power consumption of this equipmentis therefore an indicator of the increased demand for computingpower, as mentioned above. Reductions in IT electrical load musttherefore be made in other ways, in addition to hardwareefficiency improvements.

An important point when considering the reduction of datacenter energy consumption is that reliability cannot becompromised. When making changes, transient conditions suchas the loss of utility power and subsequent loss of cooling untilstandby generators can be started and brought on-line must betaken into account, in addition to normal operating conditions.The need for high reliability during such transient conditions canlower the maximum energy efficiency gains that could be

achieved in the normal operating condition; however, this isunavoidable given the mission-critical nature of most data centerapplications.

MEASURING THE SUCCESS OF ENERGY EFFICIENCY

IMPLEMENTATION – THE SOFTWARE-ORIENTED APPROACH

What is often lacking in current literature on this topic is theneed for an infrastructure that allows for the successful trackingof energy efficiency measures. This infrastructure allows trackingof energy consumption as a function of time down to the desiredlevel, giving a clear measure of the effectiveness of energyreduction techniques. If energy reduction techniques are stagedso that they are implemented at different times, this infrastructuregives insight as to the effectiveness of each technique.

Such an infrastructure consists of:

1. Appropriate measuring and recording devices and theirrelated instrumentation hardware

2. Network communications between measuring/recordingdevices

3. Software that interprets the results of the observedmeasurements

The importance of the software cannot be overstated. In fact, the“interpretation” of measurements can be quite sophisticated, andmay be expanded to include not only a report of current systemstatus and key performance indicators, but also advancedanalytics, budgeting, and load forecasting. In this role, thesoftware automates many of the tasks that are necessary todocument the success of energy efficiency efforts. It alsoautomates many of the tasks that are required for continuedreliable operation of the system and planning for system changessuch as IT hardware change-outs. Using software in such a roleis referred to as a “software-oriented approach.”

STRATEGIC ENERGY MANAGEMENT

Strategic Energy Management is both a business process anda framework for establishing, monitoring and verifying results ofplans for improved energy efficiency. In its simplest form, itdefines a plan for evaluating current performance, identifyingopportunities for efficiency gains, implementing actions toachieve savings and methodologies for tracking the real results ofthose actions.


11SPRING 2008

Standardized process examples are available fromorganizations like Energy Star and the California EnergyCommission and are intended as guidelines for the developmentand communication of a corporate energy efficiency plan.

A key element to the success of any energy efficiency project isaccurate and relevant data. This data, captured and presented inan intuitive and flexible software interface, provides significantinsight into efficiency opportunities, supports the business casefor investments, and reports on the successes of the actions takento improve performance. The most powerful of such softwareinterfaces are able to mask engineering complexities andrepresent the data as energy metrics and key performanceindicators (KPIs).

The detail of these metrics is limited only to the availability ofdata to support the calculations. KPIs can be compared withinand/or across facilities. Examples include:

Energy Characterization of Load Types:

• kWh over time for IT Loads

• kW per CFM for Cooling Loads

• Watts per square foot by rack, work cell, usable floor space

• Comparisons of Processing vs. Storage vs.Telecommunications energy use over time

• Operational efficiency (energy input vs. energy to the ITprocess loads) for the entire facility or subset of the facility

Energy Characterization by Cost:

• Total energy cost per square foot

• IT energy cost, by rack, by zone, by floor

By monitoring the metrics on a regular basis, it is possible togain significant insight into historical energy performance as wellas advance indications of performance degradation. Furtheranalysis and utilization of energy modeling techniques togetherwith monitoring metrics can shift energy efficiency efforts fromreactive model to proactive management.

REAL-TIME MONITORING AND ANALYSIS

For real-time analysis, key performance data is monitored inreal-time and typically can be displayed in any way the userchooses. The real-time analysis of energy data is, in general,operating in parallel to that for power quality data, such as asequence-of-event recording system, and may or may not be partof the same software platform.

As a tool to flag gross changes in energy usage, real-timemonitoring and analysis is invaluable. It helps establish energyusage patterns for equipment and identifies equipment andsystems that are operating outside of these normal energy usagepatterns. As a piece of equipment, system, or subsystem fallsoutside of its normal energy usage pattern, action can be takenfor planned maintenance in an orderly fashion to keep the datacenter at peak efficiency. Real-time monitoring and analysis isalso crucial for the real-time control of energy.

During the initial phase of operation of a new data center, energymonitoring and management software can begin data collectionto establish actual energy use patterns of the facility and many ofits subsystems. Data cleansing algorithms can provide automaticdata correction of metered energy data and fill gaps in missingdata.

Care should be used in attempting to compare real-timevalues of the energy metrics, against their associated establishedbenchmarks. Unless the benchmark in question was specificallydeveloped for use in real-time, the results will not be meaningful.This is due to the fact that the energy used (and its cost)accumulates over time, and it is this use of energy over time,when properly kept within the strategic performance goals of thefacility, that gives the energy savings desired.

LONG-TERM MONITORING AND ANALYSIS

Long-term data can be used to provide accurate indicators ofthe energy performance of the data center. It is this data for whichthe energy metrics and their associated benchmarks are typicallydeveloped.

In long-term monitoring and analysis, energy usage informationis gathered and documented over time.

Such data includes the specific data monitored by the hardwaremonitoring layer, plus:

• Monthly electricity, fuels, water and sewer, and other utilitybills;

• Production or data throughput over time;

Sample Strategic Energy Management process diagram. Enhanced Automation - Business Case Guidebook, California

Energy Commission, 2004


6394D_MTUDD Genset ad.qxp 9/13/06 4:32 PM Page 1

Tackling Today’s Data Center Energy Efficiency Challenges: A Software-Oriented Approach

• Data related to weather, occupancy, unplanned shutdowns,and other influencing factors;

• Significant events during the base period of analysis; and

• Hardware change-outs that can affect energy usage.

BASELINING AND BENCHMARKING

Analysis is performed after long-term data has beencollected. The first step in this analysis is baselining andbenchmarking. In general, whole-building data should becollected for at least one-year prior to this process. As discussedabove, baselining determines the starting point from which tomeasure progress. Benchmarks, when applied to a single facility,compare the energy performance to peers and competitors overtime. For a group of facilities owned by one entity, benchmarkscan also be used to compare facilities to one another in order toprioritize which facilities to focus on for improvements. Inperforming baselining and benchmarking, the actual energy useover the calendar year, versus month of the year, versus day ofweek, versus time of day, etc., should be analyzed. Any periodsrelated to holidays, extreme weather catastrophes, national orinternational events, etc., should also be taken into account.

TREND ANALYSIS

Once baselines and benchmarks are established for theimportant energy metrics, data continues to be gathered overtime and analyzed in order to understand energy usage patternsand trends. Such trends can be used to help identify improvementpotential in key equipment, systems, and subsystems.

Energy use and performance patterns over time can beevaluated to:

• Determine the capability of existing equipment at presentand for future growth;

• Identify energy saving measures that can be implemented atlow cost for cost savings;

• Identify problems with HVAC or other infrastructure systems;

• Enable prompt action to repair or replace faulty or failingequipment; and

• Help identify what changes can and cannot be made due toeffects on system reliability.

Using long-term data, the software should have the ability tomodel energy use while accommodating changes in weather,production, or occupancy. It should also be able to generate anoverall model – as well as sub-models – for time ranges such asweekdays, weekends, holidays, shifts, etc. Flexibility in datacollection should allow data to be collected and modeled overvarious interval time periods including 15-minutes, hourly, daily,weekly, and monthly.

The software should also have tools available to forecast energy

use at the site based upon the energy model. A benchmark canbe established to compare one process to another or,automatically normalized for weather variations, compare onefacility to another located in a different vicinity.

Various graphical depictions of performance provide a visualmeans of evaluating performance and savings. The softwareshould have the capability to correlate energy use to weather orother influencing factors. Also, management reports should beautomated to convey information related to energy cost andperformance history.

Another useful function is the ability to provide a graphicalcomparison of the cumulative variation of actual energy valuesversus modeled baseline values. If actual consumption perfectlymatches the baseline, then this cumulative sum will be zero. Ifthe actual consumption is higher than the baseline, it willpositive. If the actual consumption is lower than the baseline, theCUSUM will be negative. The report screenshot below shows anexample:

Sample of a “Cumulative Summation” report

By assessing energy performance over time, creative solutionscan be developed to lower the overall consumption of energy.

Such solutions can:4

• Categorize current energy use by fuel type, operatingdivision, facility, product line, etc.;

• Identify high performing facilities for recognition andreplicable practices;

• Prioritize poorer performing facilities for improvement;

• Understand the contribution of energy expenditures tooperating costs;

• Develop a historical perspective and context for futureactions and decisions; and

• Establish reference points for measuring and rewardinggood performance.

12 SPRING 2008


6394D_MTUDD Genset ad.qxp 9/13/06 4:32 PM Page 1Newslink_S08_11fx.q7:Newslink_S08.q5 4/11/08 2:32 PM

14 SPRING 2008

Tackling Today’s Data Center Energy Efficiency Challenges: A Software-Oriented Approach

SUMMARY

Energy consumption in the data center is growing; with this increased consumption comesan increased price tag, both in energy prices and the capital costs associated with theinfrastructure that brings energy to the computer. Techniques for the reduction of energyconsumption in the data center environment take many forms, and if these are implementedwithout an over-all strategy the maximum over-all energy savings will not be achieved. Such astrategy involves a large amount of energy usage data and a large number of computations, someof which must be performed for different sets of data. Doing this by hand is labor-intensive andmay not give optimal results. Instead, the implementation of automated monitoring and softwareanalysis allows the gathering of energy usage data and the analysis of this data to achievemaximum energy savings. Such a system can use elements of already-installed power monitoringhardware. Investing in such a system can allow data centers to realize energy savings today, and increase energy savings in the futureas new energy-saving techniques are developed and implemented.

REFERENCES1 Koomey, J. G., “Estimating the Total Power Consumption by Servers in the U.S. and Around the World”, February 2007, available athttp://www.koomey.com

2 Brill, K., “Data Center Energy Efficiency and Productivity”, 2007, available at http://www.uptimeinstitute.org

3 Belady, C. L., “In the Data Center, Power and Cooling Costs More Than the Equipment it Supports”, Electronics Cooling, vol. 13, no.1, February 2007

4 Selected Text from the EPA Energy Star website: www.energystar.gov

For more information contact Michael Silla at 212.615.3643 or [email protected]

Cosentini MISSION CRITICAL has Engineered Millions of Square Feet of Raised Floor Space Worldwide.

New Data Center DesignData Center ExpansionData Center Power & Cooling UpgradesConsolidation Projects

MISSION CRITICAL SERVICESFeasibility & Due Diligence StudiesSite SelectionPeer ReviewFacility Evaluation & AnalysisDesign ServicesConstruction AdministrationCommissioningPost ConstructionLoad Management Analysis

James Parker, PE, CEM, CMVP is the Engineer of Schneider Electric Engineering Services, LLC. He can be reached at [email protected].

Bill Brown, PE is the Lead Engineer of Square D Critical Power Competency Center. He can be reached at [email protected].


The paradigm of requiring 15 minutes or more of battery backupfor mission critical UPS system reliability is an antiquated andflawed perception. When properly integrated and maintained,standby generators can and will reliably support the critical loadin 10 seconds or less. This challenges the idea that lead-acidbatteries and extended backup time are necessary. The growingintolerance to a “graceful shutdown” also renders 15 minutes ofbackup moot. The UPS system can be designed with muchhigher reliability and predictability by using more reliablebackup energy methods and applying proper design techniques.This paper discusses the issue and methods of implementing ashort ride-through system with higher reliability andpredictability than with traditional methods.

Some engineers and power system designers feel there is asignificant difference between typical applications of 25 to 30seconds of reserve time with a flywheel UPS system, and 15minutes in a conventional static UPS and battery system. Thereis a reliability difference, but the actual advantage may surprisemany.

HIGH AVAILABILITY DESIGNREQUIREMENTA properly designed high availability criticalpower system requires, by definition, thediesel generator will start and assume theload when commanded. If it does not, thereis no hope to achieve 99.999 percent orhigher availability. This is accomplished byoptimizing individual standby generatordesign, operation and maintenanceparameters. Fuel analysis is normallyrequired and redundant startingbatteries/circuits or other options. Rigorousmaintenance and high-criticality testingprocedures are required which are well inexcess of those deployed for conventionalstandby applications. These measures aloneincrease standby generator start reliability

levels by more than an order ofmagnitude over general dutystandby generators. For extremeavailability applications, standbygenerator redundancy isemployed. The designer mayrequire N+1, N+2 or even N+Nlevels of standby generatorredundancy, depending uponavailability goals.

Again, the base assumption must be the standbygenerator(s) will start and assume the load, firsttime, every time. This is in fact what occurs inactual practice barring flaws in design, test ormaintenance.

Martin Olsen

Standby generator

Flywheel assembly

Flooded battery

Valve-Regulated LeadAcid (VRLA) battery

16 SPRING 2008


17SPRING 2008

RESERVE TIME REQUIREMENTSGiven the above, a flywheel UPS provides more than enoughtime to employ standby generators as standby power sources. Infact, sufficient time exists to monitor utility for seconds tominimize unnecessary diesel starts and further reduce standbygenerator dependency in the availability equation. Theadditional reserve as provided by 5-minute or 15-minutebattery-based systems is superfluous. A 15-minute allowance for“soft shutdown” of computer or other loads is entirely irrelevantsince a shutdown after such times is intolerable to mostbusinesses. The standby generator(s) must start, and in fact doover 99.5 percent of the time according to the IEEE (Institute ofElectrical and Electronics Engineers) Gold Book statistics.Arguments that additional time allows for a “second crank” arealso without merit because if, in the rarest of circumstances, thegenerator system described above does not start within the firstfive or six seconds, then like a car, it will likely not start withinthe next 15 minutes either.

REDUCING EXPOSURE TO THERMALRUNAWAYThe single most effective way of reducing the exposure tothermal runaway in a data center (a condition where serversoverheat and shut down due to a power outage) is to reducethe time it takes to re-establish power to the cooling(mechanical systems). One way of accomplishing this is toreduce the time it takes to signal the standby engine to crankup. To avoid the commonly understood ghost startphenomenon, where engines are signaled to start up eventhough power is restored within seconds of the original outage,the system has to be configured appropriately. According toEPRI (Electric Power Research Institute), nearly 99 percent of all

outages are less than 10 seconds, which means the delayshould be set at 5-10 seconds. In properly designed, high-availability critical power systems by definition, the dieselengine will start and assume (walk-in) load when commanded,typically within 5-6 seconds.

A combined 10-15 seconds from the time of a power outageuntil power is restored to the mechanical system stands incontrast to the commonly specified 100-120 second time lapsein legacy power system designs today. A reduction of thismagnitude will significantly reduce the ambient temperature andthereby the risk of thermal runaway or self-protective shut downby the servers. As an example, a 10,000 watt cabinet filled withblade servers will rise 0.4619 degrees Celsius per second duringthe event of a cooling outage. If the commonly specified 120second time lapse was employed in the power systemarchitecture, the temperature rise would be 55 degrees Celsiusor far higher than the servers own maximum temperaturethreshold. However, if a much shorter time lapse of 15 secondswas employed in the same power system architecture, thetemperature rise would be reduced to 6.9 degrees Celsius, an 87percent improvement over that of commonly specified powersystem architectures.

Arguably, this would be a simple change even in existingfacilities, but an even greater point is why then the need for thecommonly recommended 15 minutes of ride through time? Itwould appear, as densities in data centers continue to grow, the15 minute battery bank is significantly overrated from a numberof standpoints:

1. 15-minute allowance for “soft shutdown” of computer orother loads is entirely irrelevant since, by definition,shutdown after such times is intolerable to most businesses.

2. 15-minute allowance for a “second crank” on the standbyengine is also without merit because if, in the rarest ofcircumstances, the engine system does not start within thefirst 5 or 6 seconds, then like a car, it will likely not startwithin the next 15 minutes either. On July 24, 2007, it wasreported that data center developer and operator 365 Main,Inc., was impacted by a power outage in their SanFrancisco, Calif., data center and the engines failed to startup. It took a reported 37 minutes to manually crank theengines back up.

3. 15-minute ride-through would have devastating effects onthe temperature in the data center. On November 12, 2007,it was reported that data center developer and operatorRackspace, Inc., was impacted by a power outage in theirDallas/Fort Worth data center. During restart of the chillers,the data center recorded rapidly increasing temperatures anddecided to turn off loads to mitigate a thermal runaway.

Ironically, high availability UPS designs do not favor battery-based systems regardless of runtime, but rather flywheels. This isintrinsic in battery construction. Even if a given battery systemstarts off with high reliability, it falls off rapidly with time. This isparticularly true with VRLA batteries, which show a documented20 percent failure rate in just a two to three year age range. Iffailed VRLA cells are not immediately replaced, even withredundant strings, the probability a complete load loss fromopen circuit failure is certainly far greater than the possibility ofload loss from any failure of the standby generator to start in aflywheel-based system.

Simplified cause and effect diagram during a power outagewith rapid engine start

Martin Olsen is Director of Product Management and Developmentof Active Power. He can be reached at [email protected].


www.cumminspower.com/local

“Six-nines” reliability

www.cumminspower.com/solutions

Total power solutions

18 SPRING 2008

As a data center professional, you are quite busy running amission critical site - the data center. You have budgetresponsibilities, customers requiring attention, compliance

issues, energy management, and greening your operations.

And oh, by the way, there are data protection issues, backups to do,data to move off-site, high availability issues, maintaining youralternate site—whether it is internal recovery or a commercial recoveryvendor—and don’t forget contract management, and schedulingtechnical test time!

Do you find yourself in asking any of the following questions?

� How can you be sure that you are protecting the right data? Do youjust grab it all just in case? Is this an educated guess?

� Are you in a position where you are backing up the critical data thatwas defined in a business analysis or data retention study severalyears ago?

� When does a new system or pilot application become critical? � When does a critical system evolve to where it is no longer critical? � Who tells you that you can turn off that application that you havebeen faithfully backing up for years and the original owner no longerworks for the company?

� Where do you find the time and resources to track down someonewho will authorize you to discontinue backups you have been doingforever?

� Are you certain that they know what they are doing?There is an old saying that there is a difference between efficient andeffective.

Business continuity planning evolved from disaster recovery planning– ensuring the recovery of data and technology once it wasinterrupted. Business continuity planning introduced the concept of

continuous operations by protecting business processes and the dailyoperations of the organization in order to minimize the impact of adisruption and to continue operations when at all possible.

Consider the following elements of a business continuity program andthe role of the data center professional:

� Technology continuity;� Workforce continuity; and� Supply chain continuity.

So what is the role of the data center professional in businesscontinuity planning?

ONE: Who is Responsible for Business Continuity Planning?

Find out WHO is responsible for the business continuity program foryour organization. Make the Business Continuity Planner your newbest friend. Find out what initiatives are going on with the BC programand make sure you are included in the planning for those initiatives sothat the data center needs and issues are represented and heard earlyon.

The BC planner facilitates the business continuity program for theorganization. The BC planner should be interfacing with the entireorganization, including YOU! Your job is to care for the data centerwhich plays an essential role in supporting organizational objectives.Make sure your voice is heard.

TWO: Ensure the Data Center Response is Integrated with theResponse of the Organization

Understand the importance of the relationship between the IT DisasterRecovery Plan and the Business Continuity Program. You shouldclearly delineate that DR is about what happens in the data center andBC is what the organization needs to survive at minimum and becomeresilient as an ideal condition.

It is essential that data center professional get out of the data centerand be a part of the organization’s business continuity planning team.Being a part of the IT Disaster Recovery team is important, but being

The Role of Data Center Professionals in Business Continuity PlanningThe Role of Data Center Professionals in Business Continuity Planning

James I. Nelson

EFFICIENT IS DOING THE JOB RIGHTEFFECTIVE IS DOING THE RIGHT JOB


© 2007 Cummins Power Generation Inc. All rights reserved. Cummins Power © 2007 Cummins Power Generation Inc. All rights reserved. Cummins Power Generation and Cummins are registered trademarks of Cummins Inc. PowerCommand is and Cummins are registered trademarks of Cummins Inc. PowerCommand is a registered trademark of Cummins Power Generation. “Our energy working for you.” is a trademark trademark of Cummins Power Generation. “Our energy working for you.” is a trademark of Cummins Power Generation. CPG-617 (9/07)

To operate with the utmost reliability, your data center To operate with the utmost reliability, your data center must achieve 99.9999 permust achieve 99.9999 percent uptime. PowerCommand®

pre-integr pre-integrated power generation systems from Cummins Power Generation Inc. help give you end-to-end rPower Generation Inc. help give you end-to-end reliability. They’re s They’re specifically designed to fit and work together —providing providing 24/7 assurance of uninterrupted power, exceptional data rexceptional data reporting and the capacity to match your expanding infrastructurmatch your expanding infrastructure.

Our network of local distributors is at your service Our network of local distributors is at your service to deliver everything frto deliver everything from design assistance and installation to 24/7 maintenance and support.installation to 24/7 maintenance and support.

Contact your local distributor today or Contact your local distributor today or visit www.cumminspower.com/local

“Six-nines” reliability

PowerCommand power systems are fully integrated for seamless transfer of power to your loads, delivering better performance with increased functionality. They combine digital paralleling systems, automatic transfer switches, remote monitoring and microprocessor-controlled generator sets.

For more information, contact your local Cummins Power Generation distributor or visit www.cumminspower.com/solutions

Total power solutions

The Role of Data Center Professionals in Business Continuity PlanningThe Role of Data Center Professionals in Business Continuity Planning


INNOVATION TECHNOLOGY QUALITY WORLDWIDE

Ready to go Wireless?

MidtronicsWillowbrook, IL USAUSA Toll-free Sales: 1.800.766.19951.630.323.2800www.midtronics.com

Phones are wireless, computer networks are now wireless, shouldn’t your battery monitoring be, too?The CELLGUARD™ Battery Monitoring System from Midtronics is wireless, which reduces installationcosts (up to 50%) and the overall cost of ownership (up to 25%).

CELLGUARD is built around patented conductancetechnology, the most widely researched batterystate-of-health measurement. The conductance,voltage and temperature of each battery in yourUPS are monitored. The often forgotten, but critical,inter-battery connections are also monitored, ensuring e�ective analysis of battery health andsystem performance.

Network-ready system management software captures and posts data for remote review andtracking, anywhere in the world. E-mail and textmessage noti�cations alert key responders when power system faults have occurred.

Visit our website at www.midtronics.com for information on how you can receive a no-obligation analysis ofyour monitoring needs so that you can “cut the battery monitoring cord” and your system and maintenancecosts at the same time!

CELLGUARDCELLGUARDC ’S’ internet-based software adds the peace of mind ofproviding alerts and system analysis any place where internetaccess is available

Midtronics EcoImproving our environment through better battery management


INNOVATION TECHNOLOGY QUALITY WORLDWIDE

Ready to go Wireless?

Midtronics EcoImproving our environment through better battery management

21SPRING 2008

tied into the organization’s businesscontinuity structure is essential.

Also, ensure that the IT DR plan isintegrated with the overall businesscontinuity and crisis managementprogram. All too often, the IT DR planhas its own crisis managementstructure that is isolated and separatefrom the rest of the organization. It isessential that when the organizationexperiences an event that couldinterrupt operations that the data centerprofessionals responding to the eventare a part of the overall response effort.

Ensure that a representative from thedata center is part of the crisismanagement or incident responseteam. Also include a member of the business continuity team on the ITDisaster Recovery team.

THREE: Review Results of the Data Gathering Effort

When the organization participates in its annual data gathering effort,make sure that data center personnel are included in this process.This includes data gathering activities such as the business impactanalysis, risk assessments, vulnerability assessments, gap analysisand cost benefit analysis.

Ensure that data center personnel are included in the reading of the“initial/draft” findings of the data gathering effort. Review what thebusiness is defining as their needs and requirements and providefeedback back to the team.

Pay close attention to what is defined as the risk/loss tolerance; whattechnical services are required when and by which units. This is theclassic first pass at the two headed monster of “Recovery PointObjectives” (RPO)/ tolerable loss of transactions and “Recovery TimeObjectives” (RTO)/ maximum acceptable downtime. Oftentimes thebusiness chooses an RTO that is very difficult to achieve in the datacenter.

The BC planner can be your key source to understanding whatapplications, systems, infrastructure and technical services thebusiness is defining as critical.

You can and should guide and assist the BC planner and participatein this process to make sure you are getting the answers to you needto all of these questions. But, let the planner do the heavy lifting onthis effort.

FOUR: Be a Part of the Strategy Review Process

Strategy development is the beginning of a negotiation or iterativeprocess. This is one of those rare opportunities when the organizationcomes to the data center professional with a technical problem andyou have the opportunity to craft a solution for them.

A normal expectation is when you complete the effort of defining astrategy and solution and assign a cost to implementing this there willbe some interesting discussions. Someone needs to pay for thewishes and desires of the business. This will usually be an iterativeprocess involving the various business folks, management, datacenter professionals, and the BC planner. Being a part of this processearly on will ensure that the organization does not waste timereviewing strategies that will not work in your current data centerenvironment.

FIVE: Integrate IT Tests with Business Continuity Exercises

Ensure that all exercises and tests include both IT DR professionalsand the business continuity personnel.

It is no longer enough for IT planning and testing to be done in avacuum separate from the rest of the organization. IT planning andtesting needs to include the use of the alternate facility and includethose people who will be using the technological infrastructure you aretesting.

In order to ensure that you are backing up and restoring thetechnology needed by the organization, all exercises need to beintegrated with the business units.

SIX: Support Workforce Continuity

In order to have workforce continuity, you will need to develop theability for key employees to work remotely for more than a few hours.The ability of the organization to continue operations will depend onthe ability of the IT infrastructure and the data center specifically toprovide for employee remote access when the organization faces asustained event where people are unable to work from their usualoffice space.

In order to provide the technology necessary for workforce continuity,you will need to have a good understanding of business needs so thatyou can develop the best strategies to support employees workingfrom home or from an alternate location.

SEVEN: Ensure Supply Chain Continuity

The supply chain of the data center needs to be fully evaluated andvetted to ensure that the organization is not vulnerable due to a supplychain interruption. Take a close look at your suppliers and ask toreview their business continuity plans. Review contracts and negotiatethem to include compliance with your organization’s RTO and RPO.

Determine from the results of the vulnerability and risk assessmentand the business impact analysis how quickly you will require thosesupplies that support the mission critical operations of theorganization. Don’t let the data center be impacted by a broken supplychain.

In Summary

In order for your organization to operate 7x24x365, data centerprofessionals need to be a part of the overall business continuityprogram. Ensure that the data center is represented and a part of allelements of the business continuity program.

Understand that the business continuity planning effort is huge and isnot part of the core mission of data center management; however thedata center professional does have responsibility to support thecontinuity of operations by protecting the organization’s most missioncritical facility.

James I. Nelson is Chairman of the Board of Directors of ICOR. He can bereached at [email protected].

The Role of Data Center Professionals in Business Continuity Planning


SPRING 2008

Two of the most challenging questions associated with the strategicplanning of data center placing and migration is:

1. HOW FAR FROM THE PRIMARY DATA CENTER SHOULD THE NEWDATA CENTER BE?

2. HOW LARGE SHOULD THE “BUILDING” BE?

For critical applications and the capture of “in flight”; a primarydata center migration wants to be 40 route kilometers or 25 routemiles from the point of origination to point of execution. Given theroute loss of lefts and right turns and circuitous routing of fiberoptics down roads; forest fires and powerpoles, the radius of a synchronous site ring is17-20 Euclidean miles (point to point).

Synchronous Optical Networks (SONET), to beeffective and self heal, want to executebetween .25 and .35 milliseconds. Twovariables in successful encryption are traveldistances and number of “hand offs” of datafrom point to point.

The travel distance is often verified by mapping the fiber from pointto point or taking the Euclidean distance and adding approximately30-60% to the point to point distance. Short listed properties willtake a preliminary “ping” test and further applications test for the“one-way” and then the “roundtrip” data latency. Through thisprocess, the latency, corruption and lost data will be identified. Thecosts of new fiber installations to fulfill SONET rings vary from$75,000 to $100,000 per mile on existing poles to $117-$125per ft for soft earth direct bury and $1 million per mile for diggingthrough rock (3 ft down at 10 ft at crossings)

During this process the IT Departments of various organizations willidentify the critical from non-critical applications and thereforewhich applications and respective servers, routers, mainframes,etc. will remain relatively close to the existing or primary facility andwhich can or will be remote or virtual.

The other variable in shorter travel distances near or in urbanenvironments is the hand off or physical transfer of the wave ofencryption from one facilities-based provider to another. The wavehas to actually or physically go at the speed of light from one strandof CLEC for RBOC facilities fiber to a long hauled and back to localproviders. The optronic equipment acts as yellow lights on thetelecommunication super highway. Slowing it down, the more handoffs, the slower the encryption, the greater the risk of lost data, butalso bursting the signed forward once routing is determined.

Again, the only way to identify the latency of throughput is to test it.Due to the compression of the encryptedsigned on waves via WDM, TDM and DWDM,the difference between .25 and .45milliseconds at OC-192 DWDM and be theequivalent to (5) 3 hour movies of encryption.That would translate to volumes of singlespaced copy in the event of buy/sell recordtransfers for a financial institution ormillions/billions of financial exposure.

Even after the user has satisfied themselves ofthe travel distance where encryption can be successfully executedthe user will use multiple providers and paths to insure success.

The inherent challenge in the placing of the data center is that:

1. Optronics are getting better every year and the traveldistances are growing. One provider can provide successfulthroughput 200 route miles or 150 Euclidean miles.

2. Most users do not want to place a second facility within 25miles of a primary site due to the increased likelihood of a:

� UTILITY OUTAGE

� TELCO HOTEL/HUB OUTAGE

� ACTS OF GOD

� HUMAN INTERVENTION (TERRORISM)

RIGHT-SIZING:RIGHT-SIZING:

22


SPRING 2008

Affecting the primary facility. This invites the concept of a far awayor remote facility; however, optronics for critical applications are notquite there for most of the optronic software and hardwareproviders. Herein lays the conflict.

There is no preferred distance and no rules only guidelines. Thelikelihood of regional outages of critical services is real. Thechallenges of latency of critical applications of in flight data are justas real. Some users are separating the applications and inherenttravel distances of various business units; others will compromisethe remote site goals and stay local to reduce the capital expenseof two sites and the high transmission costs for remotetelecommunication.

The other main issue in the strategic planning of the data center ishow large shall it be based on some of the following inherentchallenges of:

A. Morse’s Law (Accelerating Demand of Bandwidth)

B. Useful life of environmentals (critical gear)

C. Existing “white space” usage

D. Velocity of growth of enterprise and maintenanceoperations

E. Data mining efforts

F. Consolidation of remote facilities (virtualization)

G. Microprocessor power (size of processingequipment)

H. Cooling methods and criteria for new equipment

I. Business Continuity Planning (other legacy primary sites)

J. Capital expense of asset

K. Operating expense of asset (Total Cost of Ownership “TCO”)

L. Benefits or incentives – one time and going forward

M. IT transmission costs

N. Utility rates

O. Sales tax

As one can see there are a host of critical drives in the decisionmaking process of how large to build the asset – cost, not being theleast important.

Flexibility and scalability seem to be tantamount design criteria toresilient and redundant. So how does one design, build a facilitythat has the right amount of Day 1 white space that does not havean inefficient amount of vacant expansion space? How does onenot over design with capital dollars wasted but ensure concurrentmaintainability of credited services?

What we often do is take a 24 month look at power, cooling, andwhite space absorption (inclusive of data mining efforts) andassign a velocity of growth for the past 6 months and forecast ourown complex polynomial equation for future growth or absorption.We incorporate anomolic changes and some headroom forpotential acquisitions and consolidations.

A few drivers that find their way in the decision makingprocess is the size of equipment and the limits of theireffectiveness. Power should not be distributed greaterthan 75 feet, cooling loses its effectiveness travelinggreater than 20 feet worth vertically or horizontall, etc.The headroom for power and cooling challenges areforcing us to a “Pod” or module design for growth. Thisallows for maximum utilization of the sizing oftransformers, UPS modules, RPPs, and UPS demands aswell as CRAC or CRAH units to cool same. The cost to

power and cool larger unused data centers are very meaningful andexpensive.

The optimal pod size at 100 watts per square foot is 17,000 sq ftand 15,000 sq ft for 150 watts per sq ft. The desirable floor void isa 3 feet, ceiling height is 11 feet, and ceiling plenum is 7 feet. Toomuch space is inefficient and hard to cool and move; too littlespace and the air cannot move freely – becomes restrictive and thedisplacement of heat is difficult.

Pod design allows for redundancy within the facility for critical appsand storage as well.

23

CURRENT CRITERIA FOR DATA CENTER MULTI-MARKET SITE SELECTION Ronald H. Bowman, Jr.


ExerTherm USA / North America www.ExerThermUSA.com599 Albany Ave, North Amityville, NY 11701 [email protected] Tel: 631-841-2300 Fax: 631-841-2305

24 SPRING 2008

Right-Sizing: Current Criteria for Data Center Multi-Market Site Selection

The challenging piece is how to right size for today withoutoverbuilding for tomorrow. How to grow in efficient modularity. Thehard part of these facilities is getting funded for the “heartbeat” orcritical infrastructure. The land cost is de minims.

We back into the right size of Day 1 footprint and power loadsbased on recent and relevant history of usage and mining andestablish levels of redundancy and concurrent maintainabilitymodels consistent with the business units or user Business ImpactAnalysis (BIA).The size of equipment (transformer, switchgear, UPSmodule size, generator size, cooling unit) become the buildingblocks of Day 1 and future scalability growth.

Current price models for data center builds follow the strategicquestion that needs to be weighed, leveled and scored.

The long term goals of the user are met with a matrix of how much,how long and what risk?:

• AUGMENTATION OF EXISTING IMPROVEMENTS AND LOAD?

• COLOCATION OF A SHARED INFRASTRUCTURE ENVIRONMENT?

• GREENFIELD BUILDS IN A STRATEGY OF COST, EFFECT PART OFTHE WORLD?

To satisfy the mow much question, we are recently drawn toestablished criteria of environmental integrity or concurrentmaintainability.

The Uptime Institute has established a recognizable description ofthe environmental to support to critical infrastructure in a whitepaper critically the improvements of electrical and mechanicaldisciplines. Tier 1 through Tier IV articulates the criteria of servicingneeds (N) and levels of redundancy or concurrent maintainability ofthe environments to support the IT infrastructure.

The cost to improve, execute the 3 scenarios in the Northeast inRough Order of Magnitude (ROM) are as follows for:

The one time or capital expense has an impact on the decisionmaking process, however, the Total Cost of Ownership (TCO)incorporates:

• REAL ESTATE ACQUISITION

• SALES TAX ON TELCO USAGE

• LOCAL CONSTRUCTION LABOR RATES

• UTILITY RATES

• LOCAL IT LABOR RATES

• UTILITY CAP EX FOR UNIQUE IMPROVEMENTS

• 20 YEAR OLD REAL ESTATE TAXES (CITY/STATE)

• TELECOM CAP EX FOR UNIQUE IMPROVEMENTS

• 20 YEAR PERSONAL PROPERTY TAXES

• CORPORATE OR COMPANY TAX

• SALES TAX ON IT PERSONAL PROPERTY

• PERSONAL INCOME TAX

• SALES TAX ON UTILITY USAGE

The TCO often is the driver for what cities/states get short listed andare engaged in the competition for the user.

Not all data center searches are dollar driven, however, the financialstory is often compelling in the site selection process.

Tier I Tier II Tier III Tier IVAugment existing 50-100,000 sq ftof white space

50 watts$650 psf

100 watts$850 psf

100 watts$2,700 psf

150 watts$3,500 psf

Colocation (up to 20,000 sqft)

$150 psf per annum w/oelectricity & setup



None available

Greenfield50-100,000 sq ftof white space

$650 psf $1,200 psf $2,200 psf $2,800 psf

Ronald H. Bowman, Jr. is the Executive Vice President of Tishman RealEstate Services. He can be reached at [email protected]


ExerTherm USA / North America www.ExerThermUSA.com599 Albany Ave, North Amityville, NY 11701 [email protected] Tel: 631-841-2300 Fax: 631-841-2305


26 SPRING 2008

What is Parallel Optics?Parallel Optics is a term used to represent both a type of opticalcommunication and the devices on either end of the link thattransmit and receive information. In traditional (serial) opticalcommunication, a transceiver on each end of the link containsone transmitter and one receiver. The transmitter on End Acommunicates to the receiver on End B, sending a single stream ofdata over a single optical fiber. A separate fiber is connectedbetween the transmitter on End B and the receiver on End A, sothe link contains two fibers commonly known as a duplexchannel.

In parallel optical communication the devices on either end of thelink contain multiple transmitters and receivers, e.g. fourtransmitters on End A communicate with four receivers on End B,spreading a single stream of data over four optical fibers. With thisconfiguration, a parallel optics transceiver can use four 2.5 Gb/stransmitters to send one 10 Gb/s signal from A to B. Essentially,parallel optical communication is using multiple paths to transmita signal at a greater data rate than the individual electronics cansupport. Parallel transmission can either lower the cost of a givendata rate (by using slower, less expensive electronics) or enabledata rates that are unattainable with traditional serial transmission.

Just as parallel transmission is fundamentally different than serialtransmission, parallel optical devices are fundamentally differentin construction than serial optical devices. Serial optical devicesemploy discrete components, such as transmitting opticalsubassemblies (TOSAs) or receiving optical subassemblies(ROSAs), and discrete optical connectors, such as the SC or LCconnectors, which are almost always grouped together into aduplex pair. These discrete components are not suitable forparallel devices. Two complementary technologies have enabledthe development and deployment of parallel optics devices:Vertical Cavity Surface Emission Lasers (VCSELs) and the MPOconnector.

Development and DeploymentVCSELs are semiconductor laser diodes that are fabricated in two-

dimensional arrays and emit energy perpendicular from the topsurface, as opposed to conventional, edge-emitting semiconductorlasers. Since VCSELs emit from the top surface, they can be testedwhile they are part of a large production batch (wafer) before theyare cut into individual devices. This in-process testing dramaticallylowers the cost of the lasers, while the array production makesthem ideal for forming the linear arrays used in parallel optics.VCSELs also have a better formed optical output than most edge-emitting lasers, enabling them to couple that energy into opticalfibers more efficiently.

The MPO is a standardized multi-fiber optical connector capableof bringing an array of fibers (typically 12, but capable of up to72) together in a similar form factor as a single SC connector. Byfar the most common application of the MPO is on 12-fiberoptical cables, and currently, allparallel optical devices use the12-fiber MPO similar to themodule shown in Figure 1. TheMPO connector has gainedwide market acceptance byenabling pre-terminatedmodular fiber systems. Thismarket acceptance gave parallel optics vendors a standardizedand affordable connectivity with a linear array of fibers thatmatched their VCSEL arrays.

Parallel optics devices hit the market in 1999, and in thebeginning, there were many parallel optics form factors andmultiple vendors. In 2000, the initial deployment of paralleloptical devices came right in the middle of the telecom bubble.The drive for more bandwidth led switch vendors, such as Ciscoand Juniper, to deploy parallel optics in their carrier class telecomrouters. The devices were used to interconnect line cards at veryhigh speeds through an all-optical backplane. Cisco’s CRS-1 routerhelped lead the activity and still uses parallel optics today.

During this time, several application committees, such as OpticalInternetworking Forum (OIF), Infiniband and Fibre Channel,anticipated the need for higher speed interconnects and wrote

Simon Cowley

Figure 1: POP4 or SNAP12 Parallel Optics Module

Parallel optics technology has been around for some time in high-performance network cabling, but until recently its rolehas been strictly behind the scenes. In the past two years, however, this technology has made a resurgence and is about totake center stage.


27SPRING 2008

Parallel Optics Changes the Direction of Network Cabling

specifications for 10 Gb/s, 20 Gb/s and40 Gb/s applications using paralleloptics. When the massive bandwidthpredictions for 2001 failed to materialize,parallel optics survived only in theinternal workings of equipment fromvendors like Cisco, Juniper and Alcatel,which were using the very short reach(VSR) protocol from OIF.

A Parallel Optics ResurgenceIn the past few years, high bandwidthapplications have become morecommon, typically deployed in end-usersites such as data centers or high-speedcomputing labs. Market leadingcompanies are now shipping platformsthat use parallel optical devices on thefront of the box. Parallel optics-basedFibre Channel and Infiniband cards arenow commercially available and beingused for high-speed storage and serverconnectivity. In addition, a number ofmedia-conversion applications usingparallel optics have emerged. Theparallel optical device has come out fromthe backplane and is gaining acceptanceas a useful interconnection technology.

The initial deployment of these devices inthe data center has been tying high-performance equipment together. In aswitching application, they link corerouters together in a kind of externalbackplane. Cisco’s 12000 series routersare capable of this kind of system-to-system link. In a processing application,they link many high-performance serverstogether in a computing grid. Platformssuch as HP’s NonStop series and IBM’ssystem p5 can use parallel opticaldevices to create high-performancecomputing (HPC) environments spanningmultiple computing platforms.

Parallel Optics Connects withInfinibandThe use of Infiniband as aninterconnection technology has increaseddramatically in the past two years, drivenby the higher bandwidth and lowerlatency of the protocol. It is common forInfiniband switches to have port latencyas low as 0.2-0.4 ms, with total linklatency of 1.0-2.0 ms. Compare that withtypical Ethernet link latency of 350 msand it reveals what is driving users toInfiniband. This interconnect is beingused as both a networking and storagetransport within the data center. It is

especiallypopular incluster or gridcomputing,but can alsobe found onmany

standard rack-mounted or blade servers.

As this Infiniband becomes more widely accepted, itis driving the use of parallel optics devices sincetraditional copper Infiniband connectivity has several

disadvantages. The traditional copper cables, like the ones shown in Figure 2, are large

Figure 2: Copper Infiniband Cables Comparedto an MPO Fiber Cable


www.enersys.com

Compared to 12V batteries theDataSafeTM 16V battery offers several distinct advantages:• DataSafe 16V UPS batteries deliver a

25% increase in power density and afull 15 minute run time for 750kVA in20% less space

• Simplified wiring with 1/2 the connections of 12V top terminated battery designs

• Scales easily to large capacity UPSsystems with 1/3 fewer parallel battery strings

• Front terminated for easy access and maintenance

• Easily retrofitted to many existing cabinet designs

• Uses proven, highly reliable valve regulated technology

Available exclusively from EnerSys, theworld leader in stored battery solutions.

Go to www.savebatteryspace.com and witness the revolution in UPS batteries.

943 7x24_Exchange 2/26/08 10:21 AM Page 1

28 SPRING 2008


(up to 0.45” in diameter, which is three times larger than thesmallest MPO cable) and are notoriously difficult to handle withtheir large bend radius. These cables are also not capable of beinginterconnected, meaning that all devices on an Infiniband networkmust be directly connected to each other. With Infiniband, it is notpossible to implement structured cabling using patch panels tomake system provisioning or administration manageable. Finally,copper Infiniband has severe distance limitations, with typicalmaximum channel distances being less than 20 meters.Fortunately, the parallel optics versions of Infiniband have none ofthese shortcomings. They use small 12-fiber cables that can spanhundreds of meters and are easily interconnected.

Two distinct versions of optical Infiniband have been adopted bythe market: the 4x and the 12x Infiniband. The 4x Infinibanddefines a four-lane parallel optical transceiver. The full 12-fiberwidth of the MPO interface is taken up by four transmit lanes onthe left and four receive lanes on the right. The 12x Infinibanddefines two 12-lane parallel optical devices, one device is a 12-lane transmitter and one a 12-lane receiver. The 12x applicationhas a higher potential data rate than the 4x. For example, using2.5 Gb/s lanes, the 4x device can provide 10 Gb/s, while the 12xpair can provide 30 Gb/s (but requires the use of four devicesinstead of two). Both 4x and 12x have their own specific polarity.

Media Converters Now Going the DistancePerhaps the most unexpected application of parallel optics toemerge has been its use in media converters. A media converter isany device that converts a signal from one media (e.g. coppercable) to another (e.g. optical fiber) without modifying theunderlying signal. Two applications for parallel optics devices asmedia converters have reached the market, and at least one showsgreat promise.

The first application addresses the drawbacks of traditionalInfiniband cabling while supporting the installed base of copperInfiniband network cards. Thissmall media converter, shownin Figure 3, has a pinnedcopper Infiniband connectoron one end and an MPOreceptacle on the other. Thedevice is plugged into thecopper port on a traditionalInfiniband network card andconverts the electrical signalinto a parallel optical signal. Twelve-fiber MPO cables are thenused to interconnect this first media converter to the second mediaconverter, which is attached to the traditional Infiniband copperport on the destination network card. The electrical Infinibandsignal is converted to an optical signal, transported across the fibercabling and converted back to an electrical signal at thedestination. Of course, two media converters must be used oneach link.

This application of media conversion has several compellingadvantages. First, it extends the range of the Infiniband linkexponentially, from 20 meters to 400 meters. While 400 may beoverkill for most installations, the extension beyond 20 meters is apowerful driver for the use of these devices. Second, there is a

significant reduction in cable volume. Converting the rigid 12millimeter- (0.45”) diameter Infiniband cable to a flexible 3 mm(0.11”) fiber cable reduces cable volume more than 16 times itsoriginal amount. Finally, the device converts a specialized, multi-pin connector that cannot be interconnected into a standard fiberconnector. Before, each device had to be connected directly toanother device. Now, the link can be interconnected and cross-connected, a critical part of managing the infrastructure viastructured cabling.

The second media converter application, as shown in Figure 4,converts 12 1 Gb/sEthernet signals from RJ-45 (copper) connectivityto two MPO fiberconnections. Similar tothe Infiniband converter,the fiber media converterboasts link extension andcable volume reduction.However, it providesextension where it’sunnecessary. The device

increases the link distance from 100 meters to 300 meters-plus,but 100 is an adequate distance for most applications, especiallyin the data center. The extension would be useful for applicationslike remote IP cameras, but they require much less port densitythan the 4U, rack-mounted device provides. Converting the 12 6.5mm- (0.255”-) diameter UTP cables to two 3 mm (0.11”) fibercables reduces cable volume more than 14 times, but this can bemisleading. The media converter requires that one fiber link bepresent for each copper link. If a small access switch were usedinstead, the 12 UTP channels would share just two duplex fibercords. In this case, the media converter seems to waste fiberbandwidth and require more fiber cabling than needed.

Staying PowerThe acceptance of parallel optical systems has been assisted bythe development of a structured cabling system designed tosupport this traffic. Cabling systems designed to support duplex(serial) connectivity are not guaranteed to support parallel opticssystems. The polarity requirements of parallel systems aresignificantly different than those for duplex systems. In 2005, theTelecommunications Industry Association addressed the problemby publishing TIA-568B.1-AD7, which provided guidance forimplementing MPO connectivity in structured cabling. Thisconnectivity, as shown in Figure 5, is primarily concerned withsupporting regular duplex connectivity. However, Method B withinthis standard was specifically designed to enable the cabling plantto support parallel optics applications with no change to thebackbone cabling infrastructure.

While it has taken us some time to realize the bandwidthpredictions of the late 1990s, network speeds continue toincrease, and it will not be long before 10 Gb/s is commonplace,with 40 Gb/s and 100 Gb/s defining the new high-speedapplications. The IEEE is currently developing 40 Gb/s and 100Gb/s Ethernet specifications, with a projected release date of July2010. Designated as IEEE-802.3ba, these applications will likelyuse a combination of wavelength multiplexing and parallel optical

Figure 3: Infiniband CopperFiber Media Converter

Figure 4: RJ-45/Fiber Media Converter


www.enersys.com

Compared to 12V batteries theDataSafeTM 16V battery offers several distinct advantages:• DataSafe 16V UPS batteries deliver a

25% increase in power density and afull 15 minute run time for 750kVA in20% less space

• Simplified wiring with 1/2 the connections of 12V top terminated battery designs

• Scales easily to large capacity UPSsystems with 1/3 fewer parallel battery strings

• Front terminated for easy access and maintenance

• Easily retrofitted to many existing cabinet designs

• Uses proven, highly reliable valve regulated technology

Available exclusively from EnerSys, theworld leader in stored battery solutions.

Go to www.savebatteryspace.com and witness the revolution in UPS batteries.

943 7x24_Exchange 2/26/08 10:21 AM Page 1Newslink_S08_11fx.q7:Newslink_S08.q5 4/11/08 2:44 PM

30 SPRING 2008


transmission. Over the next five to 10 years, as network speedsgain another zero and Ethernet specifies the use of parallel optics

technology, we can expect to see these devices become trulymainstream.

Simon Cowley is technical director of CommScope Enterprise Solutions. He can be reached at [email protected].

Figure 5: TIA-568B Method B, duplex and parallel


32 SPRING 2008

The Spring Conference themed “End-to-End Reliability: The GreenOutlook” will be held June 1-4 at the Boca Raton Resort & Club inBoca Raton, Florida. The Spring conference will feature compellingkeynotes, concurrent and tutorial sessions, sessions on the greeningof data centers, a spectacular vendor event, and more...

Gary Howard, Senior Solutions Architect, Data Center of the FutureDomain of Intel Solutions Services will kick off the conference with asession entitled “Efficiency Driving Intel’s Data Center of the FutureStrategy”. In addition, a panel of recognized industry leaders fromIBM, HP, PG&E and the Green Grid moderated by Mike Zatz of the EPAentitled “Greening the Data Center” will explore ways to optimize yourcomputing environments to the benefit of your bottom line as well asshare insight in applying energyefficient technology, products, skillsand services to help you reduce datacenter energy consumption. In keepingwith the theme, additionalpresentations on energy efficiency andthe greening of data centers will bedelivered with topics such as:

• Citigroup — Going Green: How to Achieve LEED Certification & ItsBenefits in Your Data Center

• APC-MGE — Greening the Data Center and Your Company’s BottomLine

• Schneider Electric — Increasing Power System Performance ThroughVendor Partnering

• Kling Stubbins — Economizers: Best Practices Energy ManagementDelivers Savings

• ASHRAE — Activities to Improve Energy Efficiency of Data Centers

• EMC — Achieving Data Center Efficiencies, Eco-Responsibility andBusiness Objectives

• Dell — Blueprint forManaging Thermals in aData Center

• Digital Realty Trust — TheKeys to Developing anEnergy Efficient Data Center

• IDC Architects — Economizethe Data Center? Programming & Airflow Modeling

• Turner — Dataville Developing Cost Efficient Data Centers

In addition to enhanced programming 7x24 Exchange Internationalpresents “Surf & Turf”. Well, not in the traditional sense... but thanks to

the partners listed below, 7x24Exchange has been able to re-createthe ever popular and fun Intra Coastalboat cruise with a twist...

Guests will be treated to dinner on landfollowed by a fun filled dessert andcocktail cruise down the beautifulFlorida Intra-Coastal waterway.

So all aboard the Catalina and Caprice, two of Florida’s mostluxurious ships, for a great evening of networking and entertainment.

Special thanks to the following partners for their support of this event:

ABB, AKF Group, APC-MGE, ASCO, Chloride, ComRent, Cummins,Data Aire, Eaton, Emerson Network Power, Gilbane BuildingCompany, IBM, KlingStubbins, Kohler, LayerZero Power Systems,Mitsubishi Electric, MTU Detroit Diesel, PDI, Russelectric, SIEMENS,Starline, Structure Tone.

End-to-End Reliability:www.7x24exchange.org

June 1-4, 2008 • Boca Raton Resort & Club • Boca Raton, FL

Register before May

2nd for a chance to win

a Toshiba TheaterWide

40” LCD 1080i HDTV

Register before May

2nd for a chance to win

a Toshiba TheaterWide

40” LCD 1080i HDTV

2008 SPRING CONFERENCE SPRING 2008End-to-End Reliability: June 1-4, 2008Boca Raton Resort & ClubBoca Raton, FL

FALL 2008November 16-19, 2008JW Marriott Desert SpringsPalm Springs, CA

SPRING 2009May 31-June 3, 2009Boca Raton Resort & ClubBoca Raton, FL

SAVE THE DATE!!!

For information about sponsoringa 7x24 Exchange event pleasecontact Brandon Dolci, CMP at(646)486-3818 x108

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

For the complete Spring Conference program andregistration information please visit 7x24exchange.orgor call (646) 486-3818.

2008 Spring ConferenceHighlights


End-to-End Reliability:

INTERESTED IN PRESENTING AT THE FALL CONFERENCE?Visit www.7x24exchange.org and download the Call for Presentations

Deadline: June 20th

SUBMIT AN ARTICLE FOR THE FALL NEWSLINKVisit www.7x24exchange.org and download the Call for Articles

Deadline: July 31st

Silver Members

CORPORATE LEADERSHIP PROGRAM MEMBERS (at press time)

MITSUBISHI ELECTRICUPS Division

Media Partners

2008 SPRING CONFERENCE

Key Members


BECOME INVOLVED IN YOUR LOCAL 7x24 CHAPTER

E U R O P E

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!

INTERNATIONAL CHAPTERS INCLUDE:Arizona

Atlanta

The Carolinas

Delaware Valley

Europe (London, UK)

Lake Michigan Region

Lone Star (Dallas)

Midwest (Kansas City)

Metro New York

New England Area

Northwest (Seattle)

Rocky Mountain

Greater Washington DC

THE PREMIER ORGANIZATION SERVING THE UPTIME INDUSTRY

Visit www.7x24exchange.org and Take Advantage of These New Services Today!

MEMBER FORUM:

Post ideas...solutions...ask industry related questions! Both end users and vendors are welcome to use this discussion forum that was built with the 7x24 Exchange Missionin mind, to promote end-to-end reliability among those individuals who design, build, use and maintain missioncritical enterprise information infrastructures.

CAREER CENTER ONLINE:

7x24 Exchange Career Center is the premier electronic recruitment resource for the industry. Here, employers andrecruiters can access the most qualified talent pool with relevant work experience to fulfill staffing needs.

• POST A JOBTarget your recruiting and reach qualified candidates quickly and easily. Simply complete our online RegistrationForm and start posting jobs today!

• POST YOUR RESUMEWhether you’re looking for a new job, or ready to take the next step in your career, we’ll help you find theopportunity that’s right for you.


35SPRING 2008

Member Advertising Rate Card

NEWSLINK OPPORTUNITIESFollowing are the Editorial Guidelines for Newslink 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 email [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 ofthe same specification, in the next or similar publication, if any proven or admitted errors or omissions have occurred. Payment is due upon receipt of the invoice. Interest shall be charged at 2% permonth compounded to yield 26.82% per year on overdue accounts. Revisions to previously submitted ad copy are subject to additional charges. A charge of $30.00 will be levied for returned checks.In the event of a contract cancellation, the advertiser or agency agrees to repay 7x24 Exchange any discounts granted for multiple insertions less any discount applicable for the number of insertionscompleted in the contract. All cancellations must be received 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

Full

Page

Ble

ed1/

4 Ve

rtica

l

1/4

Hor

izon

tal

1/3

Verti

cal

1/3

Hor

izon

tal

1/8

Verti

cal

1/8

Hor

izon

tal

1/6

Verti

cal

1/6

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

PMS Colors (add per color) $600

Process Colors (add per color) $500

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 GUIDELINES FOR NEWSLINK

Manuscript specifications: Feature articles vary in length from 500 to 2,000words. While Newslink accepts articles in a variety of formats, it prefers toreceive materials 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. Newslink doesnot pay authors for contributions.

Visuals: Authors are encouraged to submit photographs and charts, graphs, orother illustration that will help readers understand the process being described,though it does not guarantee that visuals will be used with the article. Submitall charts, graphs, and other artwork separately; do not incorporate them in thebody of the article. Indicate caption material separately. Newslink reserves theright to publish submitted visuals.

Editorial proceduresAll articles are reviewed for suitability. Accepted materials are then edited forgrammar and to conform with Newslink’s editorial style. All attempts are madeto preserve the author’s writing style, however, Newslink has the right to edit forstyle, clarity, and to fit space allotments, and to make final selection onheadlines, subheads, and graphic treatment. Manuscript submission impliesauthor agreement with 7x24 Exchange’s Editorial Policies.

CopyrightNewslink requires first serial rights for submitted articles. This means theauthor(s) grant Newslink the right to publish the article for the first time. Wealso request permission 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.


NDSL Ltd.,Oak�eld Industrial Estate,Eynsham, Witney OX29 4TSEngland

Phone: +44(0) 1865 884288

NDSL Inc.5800 McHines PlaceSuite 120Raleigh, North Carolina 27616USA

Phone: +1 919 790 7877

GET SERIOUS ABOUT YOUR CRITICAL BATTERIES, GET CELLWATCH.Cellwatch Battery Monitoring Systems

VACATION

CheckBatteries

CELLWA

TCH

INSTALLED

!

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

CheckBatteries

www.cellwatch.com

Cellwatch - We do it daily.


tipson creatinga sustainable datacenter · the time to ﬁ nd out more by contacting your pdi sales...

Documents