Cisco Cisco UCS B420 M4 Blade Server Guide D’Information
“I firmly believe that this in-house IT
system reform project will become an
exemplary model of cloud-platform
development in the industry. The
expertise we gained here is also being
applied to our BIZXAAS cloud business.”
system reform project will become an
exemplary model of cloud-platform
development in the industry. The
expertise we gained here is also being
applied to our BIZXAAS cloud business.”
Mr. Makoto
Yoshioka
Senior Executive
Manager
System Engineering
Business Unit
System Platforms Sector
NTT Data Corporation
Manager
System Engineering
Business Unit
System Platforms Sector
NTT Data Corporation
eliminate network complexity. Cisco UCS was designed with this very much in mind,
so complexity is eliminated by employing fabric-extender technology, in which the
abovementioned input/output integration and management points are greatly reduced,
facilitating network interoperability.
so complexity is eliminated by employing fabric-extender technology, in which the
abovementioned input/output integration and management points are greatly reduced,
facilitating network interoperability.
The third advantage of the architecture is its pronounced affinity with virtualization
technology. The Cisco UCS hardware was designed to maximize the benefits of
virtualization technology. For example, it has a virtual interface card (VIC) for controlling
the QoS bandwidth for each virtual NIC, and scalability allowing up to 6,400 physical
cores (as of August 2011) to be integrated in a single layer 2 network. Additionally,
using the service profile function, the operational burden of installing and swapping out
servers can also be significantly reduced.
technology. The Cisco UCS hardware was designed to maximize the benefits of
virtualization technology. For example, it has a virtual interface card (VIC) for controlling
the QoS bandwidth for each virtual NIC, and scalability allowing up to 6,400 physical
cores (as of August 2011) to be integrated in a single layer 2 network. Additionally,
using the service profile function, the operational burden of installing and swapping out
servers can also be significantly reduced.
Nor is that the end of it: Mr. Yamada also points out that the ability of Cisco UCS to
keep down the TCO of the private cloud environment was also a big attraction. “With
Cisco UCS, we could utilize the service profile function to reduce the number of man-
hours and the lead time required when scaling out the virtual servers,” he explains.
“And because the CPU is also powerful on its own, there was the potential for big
reductions in the number of licenses in the case of software such as Oracle, where the
license fee varies depending on the number of CPUs.”
keep down the TCO of the private cloud environment was also a big attraction. “With
Cisco UCS, we could utilize the service profile function to reduce the number of man-
hours and the lead time required when scaling out the virtual servers,” he explains.
“And because the CPU is also powerful on its own, there was the potential for big
reductions in the number of licenses in the case of software such as Oracle, where the
license fee varies depending on the number of CPUs.”
NTT Data’s decision to adopt Cisco UCS was made in April 2010. It started designing
the hardware configuration right away, and in July, the machines were brought in. In
August, the unit-testing was carried out, and in October, the product-testing of the
whole UCS solution was completed. From November 2010 to March 2011, the system
including administrative applications was tested, and from April 2011 into early May,
the system migration was carried out. Then in May 2011, several systems including
purchase management, electronic decision support and employee attendance
management went live in this environment. The migration of the human resources and
payroll management system, and the meetings support (project-specific decision-
making workflow) system, was also completed in August 2011.
the hardware configuration right away, and in July, the machines were brought in. In
August, the unit-testing was carried out, and in October, the product-testing of the
whole UCS solution was completed. From November 2010 to March 2011, the system
including administrative applications was tested, and from April 2011 into early May,
the system migration was carried out. Then in May 2011, several systems including
purchase management, electronic decision support and employee attendance
management went live in this environment. The migration of the human resources and
payroll management system, and the meetings support (project-specific decision-
making workflow) system, was also completed in August 2011.
Server-provisioning lead-times halved
TCO down by more than 50%
By performing integrated virtualization on system platforms, using
Cisco UCS, maintainability and operability have been dramatically
improved. According to Mr. Yamada, the shortening of server-
provisioning lead-times by means of the service profile function is
particularly noteworthy. “As long as we create the profile and install
the virtualized host in advance, when we add a new blade, the setup
is finished in about thirty minutes,” he says. “Of course, we still need
to install the OS, check the running, and do various tests like before,
but even with all this combined, the lead times are less than half
what they were.”
Cisco UCS, maintainability and operability have been dramatically
improved. According to Mr. Yamada, the shortening of server-
provisioning lead-times by means of the service profile function is
particularly noteworthy. “As long as we create the profile and install
the virtualized host in advance, when we add a new blade, the setup
is finished in about thirty minutes,” he says. “Of course, we still need
to install the OS, check the running, and do various tests like before,
but even with all this combined, the lead times are less than half
what they were.”
Sustaining the service level has also become easier. First, the use
of VIC now makes it possible to control the network bandwidth for
each virtual machine and also to secure bandwidth for high-priority
applications in the virtual environment. Even if there are major
of VIC now makes it possible to control the network bandwidth for
each virtual machine and also to secure bandwidth for high-priority
applications in the virtual environment. Even if there are major
fluctuations in traffic on the other virtual machines, stabilized access
is possible.
is possible.
Also, when maintenance needs to be carried out on a physical
server, this can be handled by migrating the virtual machine to
another physical server using the live migration function. Using VM-
FEX technology made feasible by VIC, the virtual machine’s network
policy can be kept even after migration, which makes things much
easier. This has done a lot to shorten lead-times.
server, this can be handled by migrating the virtual machine to
another physical server using the live migration function. Using VM-
FEX technology made feasible by VIC, the virtual machine’s network
policy can be kept even after migration, which makes things much
easier. This has done a lot to shorten lead-times.
TCO has also been dramatically reduced. Compared to the
previous system, initial investment has been reduced by 58%, while
maintenance costs have been reduced by 17%, and rack costs by
58%. The combined total reduction in TCO is over 50%.
previous system, initial investment has been reduced by 58%, while
maintenance costs have been reduced by 17%, and rack costs by
58%. The combined total reduction in TCO is over 50%.
VM
#1
VM
#2
VM template
UCS service
profile
Blade #2
Planned
additional
blade #8
Blade
slot
Chassis #1
Blade #1
Fabric Interconnect
Core switch
FC switch
Pre-defined
Pre-defined
Pre-defined
Pre-defined
Pre-defined
Pre-defined
VM
#α
VM
#β
VM
#γ
Blade
slot
Planned
additional
chassis #2
Important features of UCS FIC and
peripherals
• Configuration and wiring of core switch
• Configuration and wiring of core switch
and FC performed all at once during the
initial build
Important features of
VM template
• Standard hearing
• Standard hearing
sheet
• Use of standard VM
template
Important features of
profile
• Hardware definitions
• Hardware definitions
for servers to be made
in advance
• Settings for peripherals
also configured during
initial build
Important features of UCS blade
• Hardware information defined in advance
• Hardware information defined in advance
• ESX can also be installed in advance by
switching profiles
n
Shortening of Lead Times Using UCS
Initial
investment
▲
58%
5000
2500
0
TCO
(Tons)
▲
58%
▲
17%
Hardware
& software
maintenance
costs
Data
center
costs
Current system
50%
reduction
After changes
TCO reduction
(cumulative total over five years)
Relative to current system:
50% reduction
4,475
Current system
After changes
935
CO
2
reduction
(cumulative total over five years)
Relative to current system: -3540 tons
(
79% reduction
)
n
Comparison of TCO over a
five-year period UCS
n
Comparison of CO
2
emissions
over a five-year period