Efficiency Without Compromise: Optimizing Data Center Infrastructure to Reduce Cost and Deliver High Availability
Emerson Network Power: The global leader in enabling Business-Critical Continuity Emerson Technologies
© 2010 Emerson Network Power
Uninterruptible Power Power Distribution Surge Protection Transfer switching DC Power Precision Cooling High Density Cooling Racks Rack monitoring Sensors and Controls KVM Real-Time Monitoring Data Center Software
Emerson Network Power – An organization with established customers
Š 2010 Emerson Network Power
Ron Bednar Director, Marketing and Customer Insight Emerson Network Power
Š 2010 Emerson Network Power
Top data center issues
Rank
Fall 2006
Fall 2007
Spring 2009
1
Heat Density
Heat Density
Heat Density
2
Space/Growth
Power Density
Energy Efficiency
3
Power Density
Energy Efficiency
Monitoring
4
Monitoring
Availability
Availability
5
Availability
Space/Growth
Power Density
Source: Data Center Users’ Group
Š 2010 Emerson Network Power
Most energy efficient UPS solution
Š 2010 Emerson Network Power
Top data center issues
Rank
Fall 2006
Fall 2007
Spring 2009
Fall 2009
1
Heat Density
Heat Density
Heat Density
Availability
2
Space/Growth
Power Density
Energy Efficiency
Monitoring
3
Power Density
Energy Efficiency
Monitoring
Heat Density
4
Monitoring
Availability
Availability
Energy Efficiency
5
Availability
Space/Growth
Power Density
Power Density
Source: Data Center Users’ Group
Š 2010 Emerson Network Power
Efficiency considerations By making reducing energy consumption in my power and cooling infrastructure, did I: • • • • •
Increase the chance of downtime? Increase cost to run my IT equipment? Increase maintenance costs? Affect future flexibility? Impact the ability to manage my infrastructure?
© 2010 Emerson Network Power
Compromising availability for energy efficiency • Compare energy costs of two data centers 500 kW of IT, 10 cents / kW hr
Data Center A
UPS Effeciency Cost to Power the Load
92% $35,040
Data Center B Energy "Efficient" 95% $21,900
Energy Savings Per Year -$13,140
• Data Center B saves $13K per year in energy costs – UPS running on pure bypass
• Result: Improved efficiency at the expense of availability – Pure bypass = IT susceptible to surges and sags
© 2010 Emerson Network Power
Cost of downtime • Typical hourly cost of downtime by industry – – – – –
Brokerage service: $6.48 million Energy: $2.8 million Telecom: $2.0 million Retail: $1.1 million Health care: $636,000
Sources: Network Computing, the Meta Group and Contingency Planning Research
• Don’t save $13K, only to cost you and your customers millions of dollars in lost revenue – A single minute of downtime can wipe away all energy savings
© 2010 Emerson Network Power
Better to equate efficiency to lifecycle of a data center Ability to produce an output with a minimum of effort, expense, or waste
Design & Deployment
Management & Planning
Operation
Š 2010 Emerson Network Power
Efficiency Without Compromise
TM
“Optimizing data center infrastructure to reduce cost and deliver high availability”
© 2010 Emerson Network Power
Uninterruptible Power
Economization
Balancing high levels of availability and efficiency
Critical Services
Differences between transformer based and transformerless UPS systems Transformer
Transformerless
• • • •
• • • • • •
•
• • •
Enterprise space Dual input sources Full galvanic isolation Passive + active I/O power quality management Lower and more predictable management of fault currents Better DC fault management Reduced need for capacity paralleling; larger modules Affords best system availability
• • • •
Small/medium business space Smaller footprint; lighter weight Marginally higher efficiency Single input source No internal galvanic isolation 100% active I/O power quality management N+1: Lower initial system cost Easier expansion and change Best accommodates fuzzy growth plans Improved efficiency at high loading
© 2010 Emerson Network Power
N+1 redundancy: Too many N’s? Critical Bus Availability 1 0.995
N = 1 maximizes availability
0.99 0.985
0.98
Assumption: Single UPS + battery reliability of .99
Š 2010 Emerson Network Power
UPS systems and efficiency ranges Typical enterprise data center operating range
Š 2010 Emerson Network Power
Economizers for data centers Water-Side
Air-Side
• Pros – Can be used in any climate – Service requirements and complexities greatly reduced • Cons – Series indirect piping and control more complex – Complexity during Change-over – Risk of pitting coils if untreated stagnant water sits in econocoils
• Pros – Best in moderate climates • Cons – Ductwork required to get air to the space – Humidity control can be a challenge. Vapor barrier is compromised – Dust, pollen and gaseous contamination sensors are required – Hard to implement in “high density” applications – Mildew minimization actions required
© 2010 Emerson Network Power
Air economizer system Air Economizer Operation
Normal operation • Air pattern same as traditional under-floor system fed by perimeter cooling units
• Basic air flow is unchanged • External air is brought into the mixing box on the top of each cooling unit and exhausted through an independent air exhaust system • Intake is enthalpy controlled • Exhaust system is controlled by the air pressure in the room
© 2010 Emerson Network Power
Langan Engineering • Environmental engineering firm Elmwood Park, NJ ̶ ̶ ̶
Company experiencing rapid growth Organization implementing company-wide green initiative Area susceptible to sags and brownouts
• Replaced seven small lineinteractive UPS units with a single Liebert NX with Softscale on-line UPS • Scalable system enables company to double power capacity without adding or modifying hardware • UPS operates at 93 percent efficiency while decreasing downtime by 20 percent
© 2010 Emerson Network Power
Variable Air Flow
Variable Capacity
Adapting to IT changes for continuous optimization and design flexibility.
Intelligent Control
Variable-capacity cooling • IT loads have a large variation in cooling and airflow requirements – Virtualization, power management, new equipment
• Need to match cooling capacity with the IT load – Eliminates over cooling and improves cooling efficiency with reduced cycling
© 2010 Emerson Network Power
Variable-capacity cooling • Situation: Over-provisioned room – Ten 30 ton units, running at 70% capacity – Install Variable Speed Drives or Turn-Off Units
30 Ton CRAHs Cooling Produced (ton) Total Power Needed to Operate Units (kW) Cost to Operate Units ($0.10/kWh) Energy Savings Extra Cooling Capacity
Current Situation
Install VSD
Turn Off 2 Unit
10 300 124 $109,000
10 240 42.5 $37,500 $71,500 30%
8 240 99.2 $87,000 $22,000 13%
30%
• Variable Speed Drives save energy and maintain proper airflow © 2010 Emerson Network Power
UPS Energy Optimization Mode What is Energy Optimization Mode?
100.0% 99.0%
• Operation of critical load on bypass • Increases efficiency by idling the inverter while the load is fed by the bypass
98.0%
• If poor quality conditions are detected, unit automatically switches to full double conversion mode • May be a viable strategy with one system in 2N environments
93.0%
750kVA NXL Efficiency
97.0% 96.0% 95.0% 94.0%
92.0% 91.0% 90.0%
© 2010 Emerson Network Power
30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0% Eff
EcoMode Efficiency
Intelligent paralleling
3 Units @ 25% Load Each = 91.5% Efficiency
% Load
• Firmware intelligence • Increases the efficiency of an N+1 system by turning off redundant module inverters • Maintains user programmed levels of redundancy and distributes off-time equally between modules • Maintains battery charging through the rectifier
% Load
What is intelligent paralleling?
2 Units @ 38% Load = 93.5% Efficiency
© 2010 Emerson Network Power
24
Intelligent cooling controls Integrated Control Solution
Improved system efficiency
CRAC CRAC
CRAC CRAC
• Liebert iCom controls establish unit-to-unit teamwork control • Temperature sensor grid allows more precise airflow delivery • Allows one unit to serve multiple rows or CAC areas
FAN SPEED CONTROL SENSORS
Unit Control Sensor TEMP CONTROL SENSORS
© 2010 Emerson Network Power
Bay Area Internet Solutions • Co-location facility in Santa Clara, Calif. ̶ 83,000 sq. ft. facility ̶ 45,000 sq. ft. of raised floor space
• Innovative “box within a box” economizer design – Takes in 200,000 CFM of cool air to supplement CRAC units – Air filtered twice - in air corridor and by precision cooling units before entering data center floor – Runs 85 percent of the time
• Liebert CW CRAC units with Variable Speed EC Fans – Allows units to adapt capacity to IT load – Reduces energy consumption 10 to 30 percent
•
500 racks of cold aisle containment – 100% separation of hot and cold air – 500% more cost effective to build than hot aisle containment
• Anticipated cooling infrastructure energy savings of 4 million kWh per year and PUE of 1.4 while achieving 99.999% availability • Awarded maximum incentive amount from Silicon Valley Power’s Customer Directed Rebate program
© 2010 Emerson Network Power
Uninterruptible High Density Power Cooling
Intelligent Aisle Containment
Delivering architectures from 1060kW/rack to minimize space and cost.
Power Architecture
Data center density increasing Drivers • • • •
Constraints Consolidation Virtualization Cloud environments
Benefits • Optimize space • Lower building costs • Energy savings in cooling system
Source: DCUG Surveys
© 2010 Emerson Network Power
Major areas of costs and savings Adding 2000 kW of IT at 20kW/rack vs. 5kW/rack • • • • •
Smaller building for high density Fewer racks for high density Capital equipment costs more Equipment installation costs higher High density cooling is more efficient Building Costs @ $250 sq ft Rack and PDU cost @ $2,500 each
Cost Difference Low Density vs High Density ($1,875,000) ($750,000)
Cooling Equipment
$320,000
Installation Costs
$750,000
Cooling Operating Costs (1yr) Total Net Savings of a High Density Design 5 yr Total Net Savings of High Density
© 2010 Emerson Network Power
($420,480)
($1,975,480) ($3,657,400)
Integrated containment • Integrates the benefits of aisle containment with the capabilities of the control for optimized cooling • Containment maximizes air utilization • The control matches cooling capacity with heat load – Variable speed fans – Variable capacity compressors
© 2010 Emerson Network Power
Savings – DX system Without Aisle Containment With Aisle Containment 75 F
89 F 54 F
With Integrated Containment
85 F
92 F
94 F
•CFM •100%
62 F
64 F
97F
•CFM •90%
62 F
80 F
85 F 62 F
62 F
54 F
CFM 67%
Without Containment
Containment
Integrated Containment
Compressor
69.7%
50.9%
50.4%
Condenser
9.3%
9.3%
9.3%
Evaporator Fan
21.0%
18.5%
7.2%
Total
100%
78.7%
66.9%
21%
33%
Savings © 2010 Emerson Network Power
High density cooling • • • •
On demand, plug-and-play flexibility to add additional capacity Cooling at the source of heat with advanced compact heat exchangers Multiple module configurations to meet any data center layout Can increase energy savings by more than 50%
© 2010 Emerson Network Power
Cooling strategies and future opportunities
Cooling System Power (kW) / IT load (kW)
Supplemental
Š 2010 Emerson Network Power
CDW Berbee •
Data center in Minneapolis – – –
• •
Hot spots throughout the data center Liebert XD system delivers cooling directly to the source – –
• • •
•
12,000 sq. ft. 600 racks 30 blade center chassis
Eight 20 kW Liebert XDH horizontal row coolers One 160 kW Liebert XDC coolant chiller
Modular cooling strategy enables future growth Achieving a consistent temperature throughout the data center Earned $1,725 utility rebate based on energy savings and efficiencies of the Liebert XD system Saved 11 kW per hour in operating costs
© 2010 Emerson Network Power
Expert Advice
Management Tools & Technologies
Improving performance of the IT infrastructure and environment
Integration & Operation Services
What is infrastructure management?
Modeling, planning, tracking and complexity reduction
Service Capabilities • Industry Leadership: Energy Logic, Emerson Green Data Center • Best practices assessments and implementation service
Visibility, Access & Contol at the rack & device
Embedded Technology Solutions
Service Processor Manager
KVM Switch
Sensor Network
Serial Console
Rack Distribution Unit
PDU
Flexible Planning and Design of Infrastructure components
© 2010 Emerson Network Power
Value, performance and confidence REDUCE RISK, ENERGY USAGE, COST INCREASE ASSET UTILIZATION and TIME TO RESPOND
Deliver value to customers by enabling visibility into the data center. Transform data into action
VISIBILITY, ACCESS AND CONTROL • Visibility to key operating parameters for power and cooling • Access creates a two way flow of data and control: information is actionable • Getting the capacity into the right space at the right time
AGGREGATE, CONSOLIDATE KEY DATA ELEMENTS • Understand the true cost of the infrastructure • Know the effects of a change • Identify the risk areas, availability, etc. • Know, Plan, and Act on the requirements
© 2010 Emerson Network Power
EXECUTE AGILE, EFFECTIVE CHANGE • • • • •
Performance Optimization Change Management (Complexity) Asset Management Maximize availability Accurate metrics to drive continuous improvement
Assessments identify weaknesses and vulnerabilities • Identifies site-specific requirements • Identifies unwanted hot spots which degrade equipment and data • Reduces operating costs and the likelihood of equipment failure and downtime • Facilitates data center capacity planning • Provides actionable information
© 2010 Emerson Network Power
Battery mean time between failure example
Battery maintenance and no monitoring experience high reliability On-site experience significantly longer runtime before a failure Serviced monitoring have experienced no outages due to bad batteries – 1.6 million run hours!
Study based on batteries under contract prior to the end of their expected service life
Š 2010 Emerson Network Power
Online university • Online university – >300,000 students worldwide • Data center – 10,000 sq. ft. – 1,000 production servers • Implemented centralized monitoring solution • Increased understanding of the current state of critical IT infrastructure through proactive monitoring and management • Data analysis and trend reporting tools allowed data center staff to analyze data and use it to prevent specific problems from recurring • Customized alerts depending on the type and seriousness of the event
© 2010 Emerson Network Power
Norton Healthcare •
•
• • •
Hospital and healthcare system in Louisville, Ky. – Rapidly growing IT organization due to implementation of electronic medical records – Complete data center renovation Liebert SiteScan providing the new NOC with centralized monitoring and control of SNMP devices through existing network management systems Staff automatically notified when environmental conditions or power capacities change Increased workload by at least 100% without increasing staff Reduced help desk calls by 20%
Before
After
© 2010 Emerson Network Power
Efficiency Without Compromise
TM
“Optimizing data center infrastructure to reduce cost and deliver high availability”
© 2010 Emerson Network Power
Q&A
Ron Bednar Director, Marketing and Customer Insight Emerson Network Power
Š 2010 Emerson Network Power
Thanks for joining us! • Register for our next Webcast on June 16, “Critical Systems Monitoring: Innovative Strategies for Effective Infrastructure Management” • Follow @EmrsnNPDataCntr on Twitter
© 2010 Emerson Network Power