Cisco Cisco UCS B230 M2 Blade Server Guia De Informação
3 © 2011 Cisco and/or its affiliates. All rights reserved.
We were very satisfied with the engagement, which included educating our team
on the intricacies of managing and optimizing the Cisco Unified Computing System.
Cisco Services provided the foundation for Avago to build its future.”
on the intricacies of managing and optimizing the Cisco Unified Computing System.
Cisco Services provided the foundation for Avago to build its future.”
Oracle Application Performance Gains
The high memory capacity combined with the Xeon CPU performance of the Cisco
UCS Blade Servers improved Oracle application performance in Avago’s environment.
“On our previous platform, memory utilization averaged 60 to 80 percent and peaked
at 100 percent,” Shah says. “Now it’s just 15 percent. Performance is so good that we
might reduce the number of servers configured for Oracle ERP when we upgrade to
release 12.”
The high memory capacity combined with the Xeon CPU performance of the Cisco
UCS Blade Servers improved Oracle application performance in Avago’s environment.
“On our previous platform, memory utilization averaged 60 to 80 percent and peaked
at 100 percent,” Shah says. “Now it’s just 15 percent. Performance is so good that we
might reduce the number of servers configured for Oracle ERP when we upgrade to
release 12.”
Business Continuity and Reduced IT Workload
With the latest Xeon processors innovations, processing is also more efficient on
the Cisco Unified Computing System. On the previous platform, average processor
utilization was 50-60 percent, peaking at 80-90 percent. Now average utilization is
10 percent, peaking at just 20 percent. “Ample processing and memory capacity
in the Cisco Unified Computing System lessens the likelihood of outages that could
threaten the business,” Shah says.
With the latest Xeon processors innovations, processing is also more efficient on
the Cisco Unified Computing System. On the previous platform, average processor
utilization was 50-60 percent, peaking at 80-90 percent. Now average utilization is
10 percent, peaking at just 20 percent. “Ample processing and memory capacity
in the Cisco Unified Computing System lessens the likelihood of outages that could
threaten the business,” Shah says.
A bonus of more efficient computing is that the IT team no longer needs to respond
to hundreds of automatically generated tickets each month. The system used to
generate the tickets when processors reached capacity, alerting staff to terminate or
redirect processes. Since moving the Oracle stack to the Cisco Unified Computing
System, tickets have disappeared entirely, freeing IT staff to focus on optimizing the
environment. High-priority, service-interrupting calls decreased by more than half.
to hundreds of automatically generated tickets each month. The system used to
generate the tickets when processors reached capacity, alerting staff to terminate or
redirect processes. Since moving the Oracle stack to the Cisco Unified Computing
System, tickets have disappeared entirely, freeing IT staff to focus on optimizing the
environment. High-priority, service-interrupting calls decreased by more than half.
More Timely Business Intelligence
Oracle users notice the biggest performance gains on long-running batch processes.
“Hours-long Oracle processes now tend to be 30-40 percent faster, which we attribute
partly to the high memory capacity of the Cisco Unified Computing System,” Shah says.
For example, the time to extract Oracle data for the data warehouse decreased from
four hours to two. Faster cycle times give Avago confidence that data extraction will not
impede performance as the data warehouse grows. Faster cycle times also give Avago
the option to extract data more frequently in the future, making sure managers and
executives can base their decisions on up-to-date information.
Oracle users notice the biggest performance gains on long-running batch processes.
“Hours-long Oracle processes now tend to be 30-40 percent faster, which we attribute
partly to the high memory capacity of the Cisco Unified Computing System,” Shah says.
For example, the time to extract Oracle data for the data warehouse decreased from
four hours to two. Faster cycle times give Avago confidence that data extraction will not
impede performance as the data warehouse grows. Faster cycle times also give Avago
the option to extract data more frequently in the future, making sure managers and
executives can base their decisions on up-to-date information.
Similarly, the time to create a daily copy of the reporting environment from the online
transaction processing (OLTP) system decreased from 30-40 minutes to 8 minutes.
transaction processing (OLTP) system decreased from 30-40 minutes to 8 minutes.
Lower Data Center Total Cost of Ownership
Ordinarily, expanding from two to three data centers would be expected to increase
operational costs by 50 percent. “Our operational costs will actually decrease by 40
percent when we expand from two to three data centers,” Shah says. A major reason
is the space, power, and cooling savings from consolidating from 84 to 30 racks.
Ordinarily, expanding from two to three data centers would be expected to increase
operational costs by 50 percent. “Our operational costs will actually decrease by 40
percent when we expand from two to three data centers,” Shah says. A major reason
is the space, power, and cooling savings from consolidating from 84 to 30 racks.
Support for Business Continuity
Finally, the fully redundant architecture of the Cisco Unified Computing System
minimizes the risk of outages that could interrupt the business. Even if a production
server fails, Avago can provision another available server in minutes and then move
the applications virtually. In Singapore, the IT team can even swap blade servers
between data centers if necessary.
Finally, the fully redundant architecture of the Cisco Unified Computing System
minimizes the risk of outages that could interrupt the business. Even if a production
server fails, Avago can provision another available server in minutes and then move
the applications virtually. In Singapore, the IT team can even swap blade servers
between data centers if necessary.
Shreyas Shah
Senior Director,
Global Information Technology
Avago Technologies