Data Center Design for the 21st Century

© 2012 IBM Corporation

David Aubuchon IBM Data Center Services Executive

July 17, 2012


Sector Geography

It represents 308 clients in 7 countries and 25 industries

Organization Size

North America 31%

Latin America

10%

Europe 29%

Asia Pacific 30%

Public 21%

Distribution 12%

Financial Services 13%

Industrial 26%

Other 18%

Comms 10%

1,000 – 9,999 employees 59%

>10,000 employees

13%

0 – 999 employees 28%

The study addressed server, storage, network, facility and data center operations to learn what is being implemented today

58% plan significant data center investments in next 2 years

68% are the first to adopt or rapid technology adopters


One in five, or 21%, of clients have highly efficient data centers—and are allocating 50% more of their IT budget to new projects

How much of your IT budget

is spent on:

35%

65%

Basic

data

centers

Strategic

data

centers

New projects

Maintain existing

infrastructure

47%

Improving data center operational efficiency allows spending to go to:

53%

• Improving public safety or

providing more citizen services

• Transforming core banking and

payment systems

• Implementing smart grids or

improving generation performance

• Allocating funds to other business

priorities

New projects


Optimize the IT and

facilities assets to maximize

capacity and availability

Design for flexibility to

support changing business

needs

Use automation tools to

improve service levels and

availability

Have a plan that aligns with

the business goals and

keep it current

Characteristics of highly efficient data centers


Design implications from highly efficient data centers

Right size capacity and

availability to get the best

return on investment

Design for flexibility to

support changing business

needs

Optimize long-term data

center operational

expenses, not just up-front

capital costs


25%

77%

33%41%

0%

14%

2%

22%

5

Basic Strategic

Right Size Data Center Facilities

Highly efficient data centers right size by matching business needs to current and future capability

Forecast space

(10-20 years)

Forecast

power demand

Implement high and low

density zones

Have fully redundant

systems


Rightsizing can yield significant benefits. An IBM data center in

Lexington gained 8x IT capacity and avoided $50M capital cost

Diagnose

Establish baseline energy use

20x more servers than needed

Consolidate and Virtualize

10x increase in asset utilization

Compute

Resources

8x growth in IT capacity

2x growth in number of servers

Improved Operations

10% improvement from less over cooling

Best

Practices

Right size infrastructure to virtualized IT workloads

Visibility, Automation,

Control

15-20% power savings from actual usage

Use analytic tools and software monitoring to right-size the facility infrastructure after server virtualization projects

Right Size


Analytical tools can be used to right-size capacity providing a solution tailored to meet your business growth

Right Size

Business unit of

hospitality organization

in low-growth market

Distribution company in

stable agricultural market

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10

Banking and finance

customer in volatile

market

IBM created a modular

expansion plan for the data

center by linking capital and

operational expenses to

business growth.

IBM recommended

retaining their existing

data center with power

and cooling upgrades,

reducing capital expense

by US$13 million.

IBM provided metrics for

a staged release of physical

space, aiding lease

negotiations and saving

US$400,000 per year.

* Based on IBM client engagements; individual results can vary.

675 percent power increase

230 percent space increase

150 percent power increase

No change in space No change in power

60 percent space decrease

Business

plan

Capacity

forecast

Design

direction

Year Year Year


Analysis shows the impact of your technology lifecycle providing improved plans to maximize capacity and flexibility for change

Today 50% of server density is < 5 kw/Rack or less

2020 60% of server density is > 10-20 kw/rack

Predicted density over a period of time is a pre-requisite to better design

Analytics can create a density requirement model to plan for changes over time

Support unknown capacity: design for 3-5x power density over a 10-year time

Designing high / low density zones saved over $1 Million per year

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

0 5 10 15 20 25 30 35

h

Kilowatts/rack

0%

10%

20%

30%

40%

50%

60%

0 5 10 15 20 25 30 35

h

Kilowatts/rack

Based on IBM estimates from client engagements

Right Size


Availability analysis can be used to identify the cost and impact trade-offs for your organization to frame a complex decision

Right Size

"N" "N+1" "2N" "N" "N+1" "2N"

3.618 3.864 3.865

Additional annual expenses2 Reduction in annual outages2

"N"

"N+1"

"2N"

Cost per outage to justify

additional expenses2

US$498K

US$765K

US$980K

Generator redundancy1

1Other applications include cooling system redundancy, electrical system distribution, UPS technology, etc …

2Based on IBM Proprietary Analytics using expected lifespan of equipment, data center vulnerabilities and operational quality

Optimal solution Determine the most

efficient design

Cost Generate data center

designs alternatives

Value Forecast for the business

benefit of each alternative

US$400M US$1M


86%

48%

68%

47%43%

27%

53%

9%

10

Basic Strategic

Design for Flexibility

Highly efficient data centers recognize that change occurs naturally and have designed more flexibility into their approach

First and fast technology

adoption

Percentage of servers

virtualized

Build in smaller

increments

Upgrade Mechanical /

Electrical equipment

without disruption


Capital and operational cost analysis shows data centers with on demand capability is a practical and cost effective solution

Data center capital costs 60% costs from mechanical /

electrical systems

Shell

7%

Mechanical

20%

Fees

24%

Fit-Up

Costs

9%

Instrumentation

& Controls

4%

Power

36%

Facility operating costs 3-5X capital costs; energy can be 75%

Cum

ula

tive C

ost of O

pera

tions (

$M

illio

ns)

Year

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Energy Cost

Staffing

Bldg. Maint. & Mgmt.

R E Tax

$100

$150

$200

$250

$50

0

Cum

ula

tive C

ost of O

pera

tions (

$M

illio

ns)

Year

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Energy Cost

Staffing


R E Tax

Energy Cost

Staffing


R E Tax

$100

$150

$200

$250

$50

0

SCALABLE MODULAR DATA CENTER ENTERPRISE MODULAR DATA CENTER PORTABLE MODULAR DATA CENTER HIGH DENSITY ZONE

Flexible Design


Data centers with a long and extended life without disruptive retrofits despite changing demands from technology is becoming a reality

0

5

10

15

20

25

30

1x 3x

• Provision for potential water cooling

• Modular UPS design and chiller plant design

• Electrical switchgear sized to grow

$M

Power Density Growth (# times)

Incremental costs

Capital Costs Provide up to 3x the power density

with lower up front capital costs

Pay now Pay later

Good design is an insurance

policy that provide flexibility.

5-10% up front capital costs

versus paying ~50% higher

costs plus a disruptive

retrofit later.

$17M $15M $14M

$25M

Flexible Design


76% 79%

49%

31%

53%

6%

17%

0%

13

Basic Strategic

Optimize Capital and Operating Costs

Highly efficient data center optimize capital and operating costs over the 10-20 year useful life of the facility

Evaluate capital and

operating costs

Measure energy

efficiency (PUE)

Follow ASHRAE industry

standards

Monitor thermal

conditions


A singular focus on capital costs can lead to sub-optimized data centers

Source: IBM Estimates

Technology #1 Technology #2 Technology #3

2

1.5

1

.5

0

Year

1 2 3 4 5 6 7 8 9 10

2.5

$3

$ M

illio

ns

Technology 1, Vendor A $2

1.5

1

.5

$ M

illio

ns

Technology 1, Vendor C

Technology 3, Vendor G

Capital Cost Analysis Show 5x difference in technology alternatives

Operating Cost Analysis Show 2.5x difference for vendor alternatives

0

Reduce data center total costs 15-30% over its useful life Improve decisions looking at vendor and technology tradeoffs

Optimize Costs


Monitoring and management tools provide on-going visibility to improve operational management and operational costs

Perforated Tiles

20%

Cable Cutouts

13.5%

Service AC Units

16.5%

Turn off

CRACs

50%

0.0

1.0

2.0

3.0

4.0

5.0

6.0

Startup Month 3 Month 6 Full Load

Insight from Analytics Power savings from improved cooling

efficiencies

On-going Monitoring 5x improvement in PUE based on

monitoring from start-up

Use real-time monitoring for day-to-day visibility Analytics provide insight to go beyond the “low hanging” fruit

PUE

Optimize Costs


“Right-size” capacity and availability

Find underutilized assets and make changes

Use analytics to define capacity and availability to meet

the business needs

Use real time management systems to stay focused

continued improvement

Design for flexibility

Design any new infrastructure investments for scalability

to support changes in demand and technology

Allow for scalability in smaller increments over time

Design for additions to be non-disruptive

Plan for water cooling to support higher density servers

Optimize total costs over the long term

Measure efficiency on a real time basis

Ensure trade-offs between capital and operating costs are

included in all business cases

Implement actions that will provide short payback in all

situations

“No regrets” actions to improve your next data center facility


A trusted partner knows one size does not fit all. They will help you determine the right solution to meet your needs

Determine the

business

needs

Map the business needs to data center technical requirements

Model outcomes

before you

implement

Model space and capacity during design process

Design a world-class data center

Design an efficient, scalable and reliable solution to meet your business needs

Increase automation

Improve operational efficiency and reduce outages due to human error

Design / build 200-300 world-class data centers each year around the world

Experts help design an efficient, scalable and reliable solution to meet your business needs

End to end solution provider from strategy to design to move in ready


Register for the data center study paper: www.ibm.com/data-center/study

Data center services: www.ibm.com/services/smarterdatacenter

For more information, please visit

http://www.ibm.com/data-center/study



http://www.ibm.com/services/smarterdatacenter