Effective data center design doesn't have to be complicated. Learn how simple topology solutions and proven, cost-effective technologies can help simplify operations and achieve the business and performance objectives of your data center.
2. Driven by Business, IT Strategy, and Customers
Multitude of Data Center Design Criteria
§ Availability
§ Location
§ Connectivity
§ Scalability
§ Energy Efficiency with 100% Availability
§ Innovation
Many Data Centers 100% Uptime is an Expectation
Simple and Cost-Effective Designs Lead to Uptime
Data Center Solutions
2
3. Uptime Institute Data and Analysis:
§ Historically vast majority of outages are caused by human error
§ In 2011 redundancy and staffing led to 67% of saves (outages
prevented)
§ Operational Sustainability reviews of dozens of live data centers in the
past 3 years
Certain Design Solutions Increase the Chance
for Human Error while other Designs Reduce the Risk
and Allow for Saves
Why Do Failures Occur?
3
4. Start With The End In Mind
Uptime Institute research and field experience confirms
§ Data center operations start before the conceptual design
phase of the project process
§ Data center operations continue to serve as the source of
continuity for knowledge management and transition to
production
Why an Operations Focus?
§ Increase return on investment
§ Increase uptime
§ Increase data center efficiency
§ Reduce costs
§ Reduce risks
4
7. Highly Operable Engineering Solution?
• Fuel Control, Pumping, and
Filtration Room
• 2(N+1) Pumps
• Complex Design was Initially
Built Incorrectly
• Operating Valves Required
Climbing on and over Piping
7
8. Highly Operable Engineering Solution
• 20,000-liter Engine Generator
Belly Tank
• Engine Generator within
Enclosure is Directly on top of
Belly Tank
• No Additional Piping or Fuel
Pumps Required
8
9. Traits of Simplicity
§ Immediately understood architecture
• 2N Topology versus Isolated Redundant
• Well marked A and B Systems
§ Avoid reliance and over provisioning of ATSs or STSs
§ Limit Complex Building Automation Systems
Allow for
§ Reduced operator error
§ Clear operations during emergency (allows for saves)
§ Reduced cording error
Simple Solutions
9
12. Design Example: 3 to Make 2 Option
• Allows the UPSs and PDUs to be loaded at maximum of 66.6%
• Requires strict management of loads and connections
12
13. Design Example: 2N Option
• Allows the UPSs and PDUs to be loaded at maximum of 50%
• Easiest layout to manage loads and connections
13
14. § Is the design characteristic a proven performance
enhancement?
§ Does the design characteristic increase exposure to
human error?
Designers, Owners and Operators must be able
to answer the following:
14
15. “Right sized” for Load
§ Key equipment designed around major components
• UPS, Engine Generators, and Chillers/Mechanical Cooling
§ Phased build out to truly allow for “pay-as-you-grow”
but without impacting the critical load
§ Modern equipment offers efficient performance across the load profile
Avoid Overspending
§ Over provisioning of ATSs and STSs
§ Excessive Redundancy at transformation layer
§ Too many system ties
Cost-Effective Solutions
15
16. Utilize Proven Technology
§ Early adopters pay a price penalty
§ Outage costs exceed incremental savings from new or
leading edge technology
Low Maintenance Systems
§ Lifecycle Cost - Purchase price not the only consideration
• Operations/energy cost
• Maintenance requirements
• Lifespan and overhaul timelines
Offer Resource Availability
§ Critical spare parts
§ Local, experienced technicians
Cost-Effective Solutions
16
18. Onsite Power Generation
§ N+R systems where R= 1 or 2 have proven effective
§ 1500 to 2000 kW Diesel EGs provide the most kW per dollar
UPS Systems
§ 2N configuration has proven most reliable preventing costly outages
§ Allow for high efficiency across a load profile of 10-50%
§ Most have proven ability to handle a step load
§ 1000 to 1200 kW static UPS systems provide the most kW per dollar
Critical Distribution
§ 2N UPS allows for independent critical distribution to IT devices
Mechanical System
§ Designed to fit within EG balance of power and climate conditions
§ N+R systems where R= 1 or 2 have proven effective
Example of a Cost Effective Solution
18
19. Simplified Electrical Solution
G
UPS System
Computer Room Cooling
House Power
Chilled Water Plant
Main Distribution
UPS System
Computer Room Cooling
House Power
Chilled Water Plant
U
A
A
A
A
B
B
B
Mech Distribution A
Mech Distribution B
Main Distribution B
A
B
G
Critical Distribution Bus A
Critical Distribution Bus B
Side “A”
Side “B”
19
20. § Scenario – Data Center w/Legacy Chilled Water with CRAHs or CRACs
§ Goal – Improve Efficiency of CRAHs/CRACs
§ Options – Replace or Retrofit with VFD or Electronically Commutated
Fan Assembly (EC)
Cost Effective Example - CRAC/CRAH Retrofit
20
21. Neither options change capacity of the cooling unit
Both technologies proven
Both options require control modification
Cost
• Initial cost lower on VFD upgrade using existing
squirrel cage fans
• EC fans offer improved fan performance
• EC fan retrofit has significantly reduced maintenance
Result – EC fan assembly had 20-50% reduced
fan energy, reduced maintenance and reduced
maintenance related risk
CRAC/CRAH Retrofit - Analysis
21
22. Low Complexity and Low Cost may lead to high
availability
§ Start with the End In Mind
§ New yet proven technologies are efficient and may
be cost effective
Reduced complexity, simplifies the design
and lowers component count,
lowers cost and reduces the risk of human error
Summary
22