Data center operational efficiency

IBM Global Data Center Study i
Data center operational efficiency
best practices
Enabling increased new project spending by improving data center efficiency
IBM Global Technology Services
Research Report
Findings from the IBM Global Data Center Study

ii Data center operational efficiency best practices
Data center operational efficiency best practices: Enabling increased new project spending
by improving data center efficiency is an IBM study that developed a data center operational
efficiency model for assessing the capability levels of today’s data center and describes the ways
IT organizations can progress along the path of data center transformation. The report was
written by IDC, which also executed the survey and interviews on behalf of IBM.
The authors would like to give special thanks to the following individuals for their assistance and
support in developing this report:
● Dr. Ian Stewart, Director of Advanced Computing, University of Bristol
● Antonio Buratti, CIO, ABI (Associazione Bancaria Italiana)
● Pierre Debagnard, General Manager of Albiant-IT, BPCE Group
● Xiao Xiao Bin, IT Manager, INESA Information Solution Group Co. Ltd
● Martin Constant, Corporate Director of Information Technology, NORAMPAC

IBM Global Data Center Study iii
Executive Summary
Today’s data center is changing rapidly. Many enterprises
are integrating new technology solutions to modernize and
evolve their organizations. Most are pursuing a path to ensure
appropriate levels of IT service delivery and cost efficiency
and alignment to business goals. For some data centers this
means providing state of the art levels of availability, flexibility,
and scalability, while for others the goal may be to provide
“sufficient” levels of services while keeping new capital
expenditures to a minimum.
Either way, data centers can be placed on a spectrum of
efficiency and flexibility. IBM and IDC have developed a
data center operational efficiency model for assessing the
capability levels of today’s data center and describing the ways
IT organizations can progress along the path of data center
transformation. There are four key stages that describe the
typical evolution of a data center as it relates to efficiency:
Basic, Consolidated, Available and Strategic.
Data centers that are operating at the
highest level of efficiency allocate 50 percent
more of their IT resources to new projects.
Figure 1: Data centers that operate at the highest level of efficiency allocate
50 percent more of their IT budgets to new projects than those operating at the
lowest efficiency level.
Applying the results of a January 2012 global study of CIOs
and IT managers to the efficiency model, 21 percent of today’s
data centers—about one in five—have reached the peak of
efficiency and are operating at the highest level.
Basic data centers
New projects
35%
Maintaining existing
infrastructure
65%
New projects
53%
Maintaining existing
infrastructure
47%
Strategic data centers

iv Data center operational efficiency best practices
Improving data center efficiency can yield tangible benefits to
the organization. This study found that Strategic data centers
were able to deliver:
● Greater investment on strategic initiatives. Staff spend more
than half of their time on new projects versus maintaining
the infrastructure, compared to only 35 percent for Basic
data centers (Figure 1). Further, 39 percent are planning
transformational projects to reengineer their IT service
delivery as compared to 23 percent.
● Greater efficiency. They enjoy more than 2.5 times the
staffing efficiency, averaging 27 servers per administrator
compared to 10 for Basic data centers.
● Greater flexibility. More than half of the companies support
a high rate of organizational change compared to just 6
percent for Basic data centers.
There were four distinguishing characteristics of companies
that have moved toward a more strategic approach:
● Optimize the server, storage, network 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.
While the “right” solution cannot be dictated by a single,
standardized blueprint, and reaching the Strategic efficiency
level may not align with the goals of all organizations, many
IT professionals are looking for something analogous to
a playbook that provides context for designing an
appropriate strategy.
About this study
The information for this paper came from a global survey of 308
IT executives in seven countries to understand the current state
of their data center operational efficiency—processes, tools,
and technologies—across eight separate areas: data center
operations, facilities management, servers, storage, network,
applications and tools, governance and staffing. The survey
was supplemented by in-depth interviews with IT managers and
CIOs from North America, Europe and Asia. For additional study
details, see Study Methodology.

IBM Global Data Center Study 1
Contents
1 Defining the state of the data center
3 Distinguishing characteristics of a Strategic
data center
8 Recommended investments to improve your data
center operational efficiency
11 Moving up the efficiency ladder: Case studies
14 Moving toward a Strategic data center
14 How IBM can help
15 Study methodology
Defining the state of the data center
There are two critical concepts to keep in mind when
evaluating the state of data center efficiency and alignment
with the needs of the business. First, there is no single “magic
bullet” indicator of movement from one efficiency stage to the
next. The data center environment is a compilation of servers,
storage, network systems, mechanical/electrical systems,
applications and tools, governance procedures and staff. The
only effective means to measure the efficiency of data center
operations is to take a holistic approach that considers multiple
measures across all elements. Second, the evolution of the data
center is a journey, one in which the destination may change as
the business needs change. This framework should therefore
not be considered a recipe that should be followed blindly, but
rather a playbook that should be flexibly applied based on the
individual needs of the organization.
Emerging from the survey responses were four distinctive
stages that differentiate data centers from one another as
IT organizations move toward business alignment (Figure
2). Each stage characterizes the data center based on a
combination of efficiency, availability and flexibility.
● Basic: The environment is relatively stable and is
maintained based on short-term objectives, with standalone
infrastructure as the norm. Companies gain the advantages
of server consolidation but have not implemented tools
to improve availability levels, which vary widely from
application to application and site to site.
● Consolidated: Server virtualization and site consolidation
are used to take out sizeable numbers of systems and
facilities and thereby lower capital costs. Server and storage
technologies are well utilized and the possibilities for
improving availability through virtual machine (VM)
mobility are beginning to be realized.
● Available: IT infrastructure is treated as a general resource
“pool” that can be allocated and scaled freely to meet the
changing demands of workloads and to ensure uptime and
performance while providing high rates of utilization. The
focus is on measuring and improving service levels while
building out governance procedures that capture
business requirements.
● Strategic: Widespread adoption of policy-based
automation tools lowers the manual complexity of the data
center and ensures availability requirements and dynamic
movement of applications and data. Instrumentation and
metrics are consistently used to validate compliance with
governance polices.

2 Data center operational efficiency best practices
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
Figure 2: The four distinctive stages of data center maturity are based on a combination of efficiency, availability and flexibility.
Operations
11+ years Data center age < 3 years
2.5+ Power usage effectiveness < 1.5
Facilities
None Mechanical/electrical redundancy Full
High Mechanical/electrical upgrade disruptiveness Not at all
Servers
<10% Percent virtualized 60%+
0-4 Virtual machines per physical server 8+
Storage
10-20% Storage virtualization 80-90%
Backup
as archive
Archiving
eDiscovery
data mapping
Network Days Disaster recovery time Seconds
Applications
and tools
Managed by
individuals
Application portfolio decisions
Centralized
approach
Governance
Slow
and difficult
Ease of decision making Quick and easy
Staffing
0-10 Physical servers per FTE Over 100
< 30% Staff time on new projects 60%+
0/44/100/0
0/9/80/0
0/44/100/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
0/9/80/0
0/44/100/0
Consolidated
0/9/80/0
0/44/100/0
Available
0/9/80/0
0/44/100/0
Strategic
Not efficient Highly efficient
0/9/80/0
0/44/100/0
Basic

Distinguishing characteristics of a
Strategic data center
With almost 60 percent of the respondents indicating plans
to upgrade their data centers in the next two years and 68
percent indicating rapid technology adoption, it is useful to
understand the key differences that characterize Strategic
data centers. Not all organizations have environments that
require data centers built to support high rates of change,
and some may never require the near-real time flexibility
and “always on” availability typified by the Strategic level.
However, for those that do require these capabilities, this
framework provides a roadmap for thinking about the future
data center and underscores how companies can build an
infrastructure that prioritizes availability and flexibility as
well as cost containment.
Companies that operate a Strategic data center consistently:
● Optimize the server, storage, network and facilities assets
to maximize capacity and availability
● Use automation tools to improve service levels
and availability
it current.
When taking a more strategic approach to data center
operations, IT organizations put the needs of the end user
at the center of their strategy. As data centers move up the
efficiency scale, many have already taken out a significant
portion of hard cost via consolidation and virtualization, and
the real focus is on providing business benefits.
These include not only application availability and
performance but, even more important, the ability to respond
rapidly to business changes. This focus on business outcomes
can result in huge payoff for organizations where revenue
generation, innovation or competitive advantage is the goal;
in comparison, efficiency and cost containment are typically
foundational elements.
Globally, the distribution of data centers follows a bell curve
with 21 percent, or about 1 in 5 data centers, operating at
the highest Strategic level of optimization, with more than
half moving into differing stages of Consolidated and
Available environments.

Moving beyond consolidation to high levels
of optimization
Consolidation through virtualization is a necessary first
step in the path to achieving data center efficiency. Most IT
organizations initially introduce consolidation into the data
center at the server level to cut costs by reducing redundancy
in physical servers. This is often followed by virtualization
in storage and networking environments, usually driven
by a similar goal of consolidation to streamline and reduce
expenditures on physical infrastructure.
Virtualization is table stakes for data center capability, and in
fact by the time data centers reach the Strategic phase they
have high levels of virtualization across their servers, storage,
and network environments and are advanced in the use of
software and automation tools. Leaders achieve significantly
higher staff productivity by managing 8.2 virtual machines
(VMs) per server, compared to 4.5 VMs per server for
Basic data centers. Key asset optimization characteristics of
Strategic data centers compared to Basic data centers include
(Figure 3):
● 48 percent of all their servers are virtualized, compared to
27 percent
● 93 percent use virtualized storage, versus 21 percent
● 92 percent use deduplication technologies, compared to
14 percent.
Figure 3: Strategic data centers are characterized by virtualization across all
components of the physical infrastructure.
48%
27%
93%
21%
92%
14%
Percent of servers
virtualized
Use storage
virtualization
Use data
deduplication
Strategic Basic

Designing for flexibility to meet changing
business needs
Change is accelerating, putting pressure on infrastructures to
keep pace. Almost 90 percent of executives operating Strategic
data centers indicate they are the first to adopt or among the
first to adopt new technology. Having a plan that is designed
to be flexible to respond to the ever changing needs of the
business and technology is critical.
Flexibility also means having the right level of availability and
redundancy to ensure meeting the service-level agreements
(SLAs). Availability and redundancy characteristics of Strategic
data centers as compared to Basic data centers include
(Figure 4):
● 47 percent can upgrade mechanical and electrical
equipment without disruption to operations, compared to
9 percent
● 90 percent have active-active configurations for their
primary data center, versus 21 percent
● 100 percent have a backup or secondary site for disaster
recovery—over half of which are hot sites—compared to
15 percent
● 46 percent take a sophisticated approach to storage
backup including synchronous replication, geo-replication
or consistency groups for multiple snapshots, compared to
8 percent
● 45 percent have a network design that flexibly supports new
services, compared to 31 percent.
Figure 4: Strategic data centers are designed with the right level of availability
and redundancy for meeting business needs.
47%
9%
Upgrade
mechanical/
electrical equipment
without disruption
Primary data
center replicated
with active-active
Strategic Basic
Replication
used for storage
backup
Network design
flexibly supports
new services
90%
21%
46%
8%
45%
31%

Employing automation tools to improve service levels
and availability
Automation is typically the next step in the data center journey.
Introducing higher levels of automation enables greater levels
of flexibility and helps support even higher levels of availability.
Greater reliance on automation tools and technologies offloads
manually intensive tasks for system administrators, reduces
error rates and ensures the performance of applications against
their SLAs. Management characteristics of Strategic data
centers compared to Basic data centers include (Figure 5):
For server management:
● 81 percent move VMs across physical hardware, compared
to 27 percent, enabling much higher levels of flexibility
and availability
● 100 percent use automation tools to manage their virtual
server environment—and 58 percent use automation
tools to move VMs automatically based on service level
agreements (SLAs), without the need of manual
intervention—versus 1 percent
● 32 percent offer a self-service portal that enables cloud-like
capabilities, versus 4 percent, and another 48 percent plan
to offer one in the next 12 months—meaning 80 percent
expect to offer one by 2013.
For storage management:
● 85 percent have automated tiered storage, versus 12 percent
● 87 percent use a service catalog approach for storage,
leading to higher levels of availability and automation,
versus only 3 percent.
For network management:
● 60 percent use automated network management, compared
to 20 percent
● 30 percent versus 3 percent use policy management
processes to automatically provision network services,
which drives faster response to service as well as network
recovery times in minutes and seconds instead of hours
and days.
Figure 5: Extensive use of automation across server, storage and network
management enables the high level of availability and service levels that
characterize Strategic data centers.
58%
1%
Move VMs
to meet
SLAs
Implemented a
storage service
catalog
Strategic Basic
Network services
automatically
provisioned
Monitor
thermal
conditions
87%
3%
30%
3%
31%
0%
For facilities management:
● 31 percent use software tools to monitor thermal conditions
versus 0 percent, providing insight to adjust to real-time
operating conditions.

Having a plan that aligns with business goals and
keeping it current
Organizations with Strategic data centers are far better
prepared to take advantage of market opportunities as the
economy rebounds. They have heavily utilized consolidation
projects to optimize the number of data center sites they
maintain. Moreover, we find they are significantly more likely
to continually evaluate the target number of data centers they
should have in a continual focus on strategy and execution.
For example, IT organizations with Strategic data centers
were much more likely to have expanded and modernized
their capabilities during the recent economic downturn. Over
60 percent of Basic data center operations made no strategic
changes or investments over the past two years, and more than
70 percent don’t expect to do so over the next two. In contrast,
nearly all Strategic data centers experienced some form of
expansion or growth over the past two years, and more than 80
percent of them expect to do so over the next two.
Additionally, IT organizations operating Strategic data centers
were more likely to regularly engage in formal planning
exercises. Characteristics of Strategic data center planning
that increase flexibility compared to Basic data centers include
(Figure 6):
● 68 percent plan to build in smaller increments of capacity
rather than build out all at once, compared with 53 percent
● 25 percent forecast the space needed to support the 10–20
year useful life of a data center, compared to 0 percent
● 77 percent forecast power demands, versus 14 percent
● 33 percent plan to implement low- and high-density zones
to support the varying power demands of new technology,
versus 2 percent.
Strategic data centers are in a better position and are nearly
twice as likely to pursue transformational projects. Fully 39
percent of Strategic data center managers are planning projects
over the next five years to significantly change the way they
deliver IT services to their organization, compared with only
23 percent of Basic data center managers.
Figure 6: Managers of Strategic data centers enable alignment with business
objectives by engaging in regular forecasting and employing expansion
strategies that ensure flexibility.
68%
53%
Build capacity
in smaller
increments
Forecast space
(10 to 20 years)
Strategic Basic
Forecast power
demand
Implement high- and
low-density zones
25%
0%
77%
14%
33%
2%

85 percent of Strategic data center managers
planning major projects, and 77 percent of
those planning projects overall, say they will
turn to outside help.
Data center managers across the board understand that this
will require investment in outside tools, technologies and
assistance. Of those planning a major project, over three in
four said they will turn to outside help, with an even greater
percentage (85 percent) among Strategic data centers. What’s
more, Strategic data centers are significantly more likely to
leverage more off-premises capabilities—including hosting,
collocation and alternative sites for disaster recovery—while
still having 64 percent of their capacity on-premises.
Recommended investments to improve
your data center operational efficiency
While cost containment is certainly a critical benefit of data
center efficiency, probably the most important benefit comes
in the ability to better serve the needs of the core business and
respond to shifts in market demand. Evidence shows that the
journey toward greater levels of data center efficiency requires
significant changes to the organization’s tools, technologies
and processes.
Once companies understand what stage they are at and
where they want to go on the efficiency spectrum, they
will need to consider appropriate investments in time and
resources. Because there are dependencies required to move
from one stage to the next, getting to the Strategic level
requires a succession of steps. Leveraging the distinguishing
characteristics of leaders applied to each discipline area can
help determine how to get started, including:
● Data center operations and facilities management
● Servers
● Storage
● Network
● Business resilience
● Governance, including applications, tools and staffing.
Data center operations and facilities management
Strategic data centers pay careful attention to their facility
design and understand the need for a holistic view that treats
the data center as a single system. They plan to meet the
business needs over the useful life of a facility by forecasting
power, space, capacity and availability—which leads to better
predictability and lower disruption during changes and build-
outs. Insights from leaders include:
● Right size capacity and availability. Forecast capacity
and availability to meet the business needs of primary and
backup centers. Then continue to manage for efficiency
using real-time monitoring and management software.
● Design for flexibility. Ensure that investments provide the
scalability required to support rapid changes in demand
and technology by, for example, building new capacity in
smaller increments over time and designing mechanical/
electrical systems so that equipment changes can be made
without disrupting operations.
● Optimize total costs over the long term. Ensure trade-
offs between capital and operating costs are included in
facilities design, and measure energy efficiency and power
consumption on a real-time basis.

“Building a new data center has provided
us with more space, more efficient and green
cooling and power, and more robust service
delivery. Incorporating greater levels of
redundancy was a key component of that.”
—Martin Constant, Corporate Director of Information Technology, NORAMPAC
Paper and Packaging industry, Canada
Servers
Strategic data center operators have consolidated their server
infrastructure to achieve the fundamental efficiencies from
server management. They understand the need to tackle the
“harder” projects to leverage automation and software tools
to drive higher levels of availability and improved quality of
service. Specific insights to leverage include:
● Move beyond consolidation to virtualization. Improve
SLA performance by using software tools and automation
to move virtual images between physical servers and data
centers based on policies.
● Prepare for cloud computing. Plan to use a self-service
portal that allows VMs to be automatically ordered online,
with a choice of size, operating system and service level.
● Take advantage of the latest technology. Knowing how
to optimize systems and move workloads will allow you
to take advantage of converged infrastructures (server,
storage and networking systems that are sold together with
management software in a pre-integrated package).
Storage
Strategic data centers are dealing with all aspects of storage
optimization and management. They realize the need to
get ahead of the explosion in storage by using software and
policy-based management systems to reduce the hands-on
labor required to provision and manage storage. Insights from
leaders include:
● Increase storage optimization. Leaders implement
four to six times more storage optimization techniques,
including virtualization, deduplication, thin provisioning
and others.
● Reduce the time spent by storage architects. Implement
storage management technologies, especially storage
service catalog, to drive self-service and policy-based
management.
● Don’t overlook storage backup and archive. With all the
focus on the volume of storage, leaders realize they need
to manage the full lifecycle of data. Consider using more
sophisticated approaches to storage backup—including
geo-replication or consistency groups for multiple
snapshots. For archiving, consider using eDiscovery data
mapping or defined processes for audits.
“We are already using automated
hierarchical storage, storage virtualization
and deduplication as well as dynamic
resource allocation (on demand).”
—Pierre Debagnard, General Manager of Albiant-IT, BPCE Group, France

Network
There are many external pressures on the network today, such
as the explosion of smartphones and how they have greatly
accelerated the demand for access to applications and data;
the growing use of video; and the adoption of cloud
computing. Leaders realize the need to have a data center
networking strategy in place. They are also moving beyond
traditional network optimization techniques to approaches
that include network management and automation in order
to improve overall IT efficiency and flexibility. Insights from
leaders include:
● Develop and execute a network strategy. Take a holistic,
long-term view that considers the network, servers, storage
and applications and end-to-end manageability balanced
against business and financial goals.
● Implement network management and automation.
Use tools and processes that enable continuous network
adjustments to meet policy-based application requirements,
and use predictive tools to avoid unscheduled outages.
● Design for flexibility. Incorporate into the architecture
the ability to automatically provision network services based
on policies, with minimal human intervention.
Business resilience
The ability to manage IT risk is essential for enabling growth,
dealing with changing business conditions and addressing
new regulations, security threats and service outages. Leaders
distinguish themselves in their approaches and their ability
to mitigate negative risks while enhancing their ability to
optimize potential opportunities. Insights from the strategies
of leaders include:
● Review your business continuity plan. Avoid reliance
on tape alone for data backup and recovery, whether at
the data center or a remote location. Combine onsite and
remote disk storage for backup.
● Understand the impact of systems not being available
to specific business processes or applications. Leaders
provide the optimal level of availability to meet business
needs by using an active-active configuration, which allows
for rapid failover of systems in the event of failure.
● Examine business and regulatory compliance
requirements. Gain an understanding of your potential
long-term data archiving needs, including how search
capabilities affect your ability to meet compliance
requirements. Have a defined process for audits and have
archive eDiscovery capabilities.
Governance, applications, tools and staffing
Strategic data center executives establish an environment that
is supportive of using a number of management best practices,
including:
● Use a centralized portfolio approach to application
management. Leaders will also apply different service and
support levels to individual applications if the application
owner is prepared to pay.
● Focus on both hard, upfront capital costs and ongoing
operational costs when deciding on data center
investments. Leaders use monitoring and management
software to ensure a total-cost focus in ongoing operations
as well as when making point-in-time investment decisions.
● Implement decision-making procedures and policies.
Leaders employ documented procedures and policies
to ease decision making regarding ongoing data center
operations.

“We currently manage applications via
individual tools, but we are planning to
implement a single management platform in
the future.”
—Xiao Xiao Bin , IT Manager, INESA Information IT Services industry, China
Moving up the efficiency ladder:
Case studies
Associazione Bancaria Italiana (ABI): Moving from Basic
to Consolidated
ABI, the Italian Bank Association, is a nonprofit organization that
represents the interests of Italian financial institutions both at
home and abroad. Headquartered in Rome, the association’s
offices are located within the Palazzo Altieri, a national historical
monument filled with significant works of art.
In 2010, ABI’s IT infrastructure consisted of 110 servers and
50 switches and routers hosted in 6 separate server rooms
dispersed throughout the building and managed by a staff of
6 IT administrators and a total of 19 IT department staff. These
systems supported 600 internal connections and several
thousand external connections through a web portal. There was
no virtualization, the server rooms were not properly equipped
or cooled, and the need to maintain each one separately led to
staffing inefficiencies. Looking to take its infrastructure to the next
level, ABI decided to push forward with consolidation, beginning
with centralizing its data center into a single facility.
One of ABI’s unique challenges was the requirement to locate the
new data center in its existing headquarters—which, as a cultural
heritage, is under significant restrictions that limit structural
modifications. Nevertheless, ABI identified a suitable space, an
860 square foot high-ceilinged hall located within the palace, and
began renovating the site.
After determining it could not completely retrofit the site (for
example, it couldn’t introduce raised-floor cooling), ABI opted
to equip the hall with an innovative cooling system based on
APC in-rack water cooling blocks connected to chillers above
the rack cabinets. By transforming its data center this way, ABI
believes it has reduced the average power consumption of the IT
infrastructure by about 35 percent, or 25 kW.
“Our system provides cooling directly to
the IT equipment, alleviating the need
to provide room-wide air conditioning.
By transforming our data center we have
reduced the average power consumption of
the IT infrastructure by about 35 percent or
25 kW.”
— Antonio Buratti, CIO, ABI
Now that ABI has modernized its facilities, with a focus on
mechanical/electrical and power and cooling, it has opened the
door to completing its transition to a Consolidated data center.
It is planning to further consolidate servers from rack-mounted
units to blades, whose higher server densities can now be
supported by the more efficient cooling system and, in tandem,
to introduce greater levels of server virtualization. In addition, it
is planning to incorporate advanced management tools that will
further increase the efficiency of the data center by enabling the
operation of the entire facility, from monitoring racks to operating
facilities security systems all from a single dashboard.

University of Bristol: Investing for a Strategic data center
The University of Bristol is a leading UK research university with
a broad portfolio of High Performance Computing (HPC)-based
studies, including climatology, aerospace, gene sequencing,
social medicine, economics and computational chemistry. Its
dedicated HPC data centers support the computationally-
intensive research and teaching needs of over 600 researchers
and students.
In addition to its main corporate data center, the University has
two separate data centers that are exclusively used to house its
HPC and research data storage systems, which are operated at
the Available efficiency level. The data center infrastructures offer
high levels of virtualization and redundancy through clustered and
distributed system configuration. The larger data center has 38
racks within an APC hot aisle enclosed solution, and the smaller
one has 12 APC racks in a similar hot aisle configuration. They
house a total of over 600 server nodes and 1.3 petabytes
of storage.
In 2006, the University devised a ten-year data center plan,
which called for a major upgrade to its infrastructure with the
addition of new data center capabilities to further increase its
levels of optimization. Unfortunately, the campus is tight on
available space, so the University came up with the solution of
transforming an old water storage facility into a new data center.
The unique space presented some unusual challenges, including
the need to move equipment up five stories as the water storage
facility was on the roof of the Physics building, and to ensure
that no electromagnetic interference affected the University’s
research radio telescope that was housed on top of the same
structure. Fortunately, it also offered advantages—such as some
free air cooling since the Physics building is on one of the highest
points in Bristol.
In the end, the new facility provided over 190 square meters
of floor space with a number of state-of-the-art capabilities
representative of a Strategic data center. These include a
modular design with two enclosed hot-aisle pods that act like
separate data centers. These APC Infrastructure hot-aisle water-
cooled enclosures can be easily scaled when the University’s
computing needs grow. The modular approach lets the University
support future scalability while saving on upfront capital costs
and avoiding overbuilding. Currently specified at 20 kW per rack,
the facility is designed to and can support higher densities in
the future.
“We have developed lights-out
administration tools that let us manage both
the data center and the HPC and research
data storage systems with a staff of four
FTEs. This alleviates the need for staff
members to enter the data center for routine
maintenance and monitoring tasks.”
—Dr. Ian Stewart, Director of Advanced Computing, University of Bristol
Further characterizing a Strategic level of efficiency, the data
center now supports a number of state-of-the-art automation
capabilities, including lights-out administration which alleviates
the need for HPC staff members to enter the data center for
routine maintenance and monitoring tasks and allows both the
data center and all the computer equipment to be managed by
a staff of four HPC system administrators. Automated scripts
communicate with APC sensor equipment to monitor the
machine room environment and take appropriate actions, all
the way up to being capable of shutting down the compute and
storage systems if something goes drastically wrong.
Looking to the future, the University is already planning to expand
its current 38 rack units to 48, targeting a completion by late
spring 2012. It is also aggressively pursuing green initiatives,
considering both the use of more power-efficient processors
and making more efficient use of that processing power through
more intelligent software. Not only will this further reduce
operating costs, but it could also extend the life of the data
center.

Albiant-IT, Group BPCE: Operating at the Strategic level
Albiant-IT is the services provider dedicated to hosting and
managing the data centers of the banking Group BPCE, a
French company offering a comprehensive range of banking and
financial services to a wide range of corporate and consumer
customers. The group has 36 million customers served by 117,000
employees and 8,000 branches. To support these operations
Albiant-IT operates a total of four datacenters across two sites,
one in the Paris metropolitan area and one in the south of France.
The four datacenters combined have a capacity of 80,800 square
feet extendable to 97,000 square feet and currently host 18,000
servers. 80 percent of the servers are x86, with the remainder
consisting of Unix servers and seven mainframes.
Albiant-IT has made strategic investments in its datacenter that
enable it to operate optimally. It operates at a five nine availability
SLA (99.999 percent uptime) and in fact has a measured 100
percent uptime since it put its current facilities in place more than
two years ago. It currently has a ten-year capacity plan in place to
account for its internal IT and hosting services infrastructure.
The datacenters operate in a hot-hot (active-active) mode with
a 2(N+1) architecture. Each datacenter replicates into the other
and workloads can be moved as necessary. Capacity can be
added in a modular fashion, both by increasing energy capacity
and by equipping new “rooms” within the existing physical
facilities. Energy consumption is measured at the facility level
according to the energy capacity plan in place. A great deal of
focus is placed on optimizing power usage effectiveness (PUE);
the facility is currently operating at a PUE ratio of 2, with the goal
of achieving 1.7 in the very near future. There is an entire system in
place to optimize energy consumption that relies on a number of
variables, including a room’s population, air flow, hygrometry and
temperature optimization.
There is widespread deployment of virtualization, with overall
server virtualization levels over 60 percent. VM movement is
supported in an automated capacity (for example, in case of
server failure). Virtualization is also incorporated into the storage
environment, with deduplication and dynamic resource allocation
on demand. Backup is performed onsite via tape and on disks
with geo-replication. The network is architected to recover
from an outage in real-time. Governance is provided through a
change committee representing each of the client organizations
within the bank, and the core decision making criteria are always
ensuring the ability to deliver high availability and minimize
operating costs.
“We operate at a 99.999 percent SLA and
have had 100 percent measured availability
since our current facility was deployed two
years ago.”
—Pierre Debagnard, General Manager of Albiant-IT, BPCE Group
Even though the datacenter is operating in most respects at
the Strategic level, this is not to say there are not future areas
of further optimization that Albiant-IT is considering. One of the
areas is the introduction of a converged infrastructure. Albiant-
IT, with its BPCE clients, is considering implementing such an
infrastructure providing it helps reduce costs and enables a quick
return on investment (ROI).

Moving toward a Strategic data center
Data centers are under constant pressure to scale and evolve
to meet the changing needs of the underlying business. To
adapt to these challenges, each data center takes a slightly
different approach.
Today, about one in five data centers operate at the Strategic,
or highest efficiency, level. Companies not yet operating at
this level can achieve greater efficiency by emulating the
four key behaviors of IT organizations that operate Strategic
data centers:
● Optimize the server, storage, network and facilities assets to
maximize capacity and availability
● Use automation tools to improve service levels
and availability
it current.
For most companies, getting there will not happen overnight.
It typically takes organizations several years of planning and
strategic investments in each area of the data center to achieve
Strategic status. While North American organizations and
companies with more than 500 employees had the highest
proportion of Strategic data centers, this level of efficiency is
achievable for any company. Strategic data centers were found
in all regions of the world and in smaller companies.
Most companies, whatever their size, plan on using outside
help with the projects that advance efficiency, a realization
especially shared by Strategic data centers. To achieve the
highest levels of efficiency, data centers must continually
re-evaluate their performance, reviewing their investments in
tools, technologies and governance, and must have the right
level of skills and assistance. Doing so can yield benefits in
greater staffing efficiencies, greater levels of flexibility and the
ability to spend more time on strategic IT initiative to support
the business.
How IBM can help
IBM helps enterprises around the world plan, optimize
and automate their data centers in order to support their
business growth and objectives. IBM has a broad portfolio
of data center facilities planning and design, cloud, IT
virtualization, network modernization, business resilience
and automation services that can help you meet your data
center efficiency objectives.
You can get started on your data center efficiency journey
by taking the Data Center Efficiency Self-Assessment. This
no-cost online tool will give you a quick snapshot of your
efficiency status across facilities management, servers, storage
and networks.
For more information
To learn more about how IBM can help you progress on your
journey to greater data center efficiency, you can contact your
IBM representative or visit the following websites:
ibm.com/data-center/study
ibm.com/services/smarterdatacenter

Study methodology
The information for this white paper came from a global survey
of 308 IT executives, conducted in January 2012, and was
supplemented by in-depth interviews with data center managers
representing each stage of data center efficiency. The survey
population consisted of IT executives who have responsibility
for or influence over their organization’s data center strategy,
from organizations of over $50 million in revenue with at least
one enterprise-class data center. Respondents were randomly
recruited and screened from international panels and came from
seven different countries: the United States, Brazil, Canada,
China, Germany, France, and India. Global data was derived by
weighting IT spending on server systems, storage, enterprise
networks, packaged software and services (excluding telecom
and outsourcing). Respondents were recruited by phone to
complete the survey over the Internet. Both phone and web
portions of the survey were administered in the local language.
The surveys asked respondents to provide information about
their data center, tools, technologies and processes across eight
separate areas: data center operations, facilities management,
servers, storage, network, applications and tools, governance
and staffing. The questions were designed to unearth the data
center’s efficiency level in each of these areas. The data from
the survey was imported into an IDC model designed to assess
and categorize efficiency in each of these areas and to roll it
up into an overall data center efficiency rating (Figure 7). The
model examines the levers by which data centers can improve
their infrastructure and identified a number of areas, including
availability/resilience, cost-effectiveness and the flexibility to
provide the capacity needed by the business.
The demographics of the respondents were:
• 60 percent from mature countries and 40 percent in
growth markets
• 63 percent from large enterprise and 37 percent from
small and mid-sized businesses
• 83 percent were IT managers and 17 percent were chief
information officers
• 25 industries covering finance, communications, industrial,
distribution, public sector and others.
The information from the survey was supplemented with five
in-depth interviews with executives responsible for data centers
in North America, Europe and Asia. The respondents had
responsibility for full data center operations and represented the
full spectrum of data center efficiency stages.
Figure 7. The study identified four stages of efficiency worldwide for
data centers.
17%
Basic Consolidated Available Strategic
32%
30%
21%
Efficiency level
1 standard
deviation
1 standard
deviation
%ofsurveyrespondents

© Copyright IBM Corporation 2012
IBM Corporation
IBM Global Technology Services
Route 100
Somers, NY 10589
Produced in the United States of America
April 2012
IBM, the IBM logo and ibm.com are trademarks of International
Business Machines Corp., registered in many jurisdictions worldwide.
Other product and service names might be trademarks of IBM or other
companies. A current list of IBM trademarks is available on the Web
at “Copyright and trademark information” at www.ibm.com/legal/
copytrade.shtml
This document is current as of the initial date of publication and may
be changed by IBM at any time. Not all offerings are available in every
country in which IBM operates.
THE INFORMATION IN THIS DOCUMENT IS PROVIDED
“AS IS” WITHOUT ANY WARRANTY, EXPRESS OR
IMPLIED, INCLUDING WITHOUT ANY WARRANTIES
OF MERCHANTABILITY, FITNESS FOR A PARTICULAR
PURPOSE AND ANY WARRANTY OR CONDITION OF NON-
INFRINGEMENT. IBM products are warranted according to the terms
and conditions of the agreements under which they are provided.
RLW03007-USEN-01
Please Recycle

Data center operational efficiency

More Related Content

What's hot (19)

Viewers also liked (9)

Similar to Data center operational efficiency (20)

Data center operational efficiency