Benchmarks of Cisco UCS
35.83 @ 26 tiles on VMware VMmark
This is a Press Release edited by StorageNewsletter.com on March 24, 2010 at 3:22 pmCisco Systems, Inc. announced that the Cisco Unified Computing System has achieved outstanding performance results in industry standard benchmarks for virtualized environments as well as other key data center applications. Of the top results Cisco achieved, the VMware VMmark benchmark has special significance for customers by demonstrating that the Unified Computing System B250 M2 Server is capable of delivering the highest performing blade server for virtualization, regardless of the number of processor cores.
These record-breaking performance results were attained with the Cisco Unified Computing System, an evolutionary new data center architecture that bridges the silos in the data center by uniting compute, network, storage access, and virtualization resources in a single, energy-efficient system. This reduces IT infrastructure costs and complexity and helps extend capital assets and improve business agility. Based on industry standards, the Unified Computing System is a key part of Cisco’s data center portfolio and Data Center 3.0 strategy.
Cisco Unified Computing System B250 M2 Server set a new record(1) with results of 35.83 @ 26 tiles in the VMware VMmark benchmark, which measures virtualization performance using a server consolidation workload. This represents a 42 percent improvement over the previous highest two-socket published result.(3) This means with the Cisco Unified Computing System, businesses can attain higher consolidation ratios in their virtualized environments with greater performance, which translates to greater return on investment, reduced total cost of ownership, and a more agile business that can deploy applications more rapidly and securely.
Cisco Unified Computing System C250 M2 Server set a new record on the SPECjAppServer2004 benchmark, which measured application server performance, running Oracle WebLogic Server application. The Cisco Unified Computing System increased performance by 30% compared to previously published two-socket single-node server results. This result translates into accelerated business productivity and agility.
Cisco also achieved top results(2) in several key High Performance Computing benchmarks including SPECfp_rate_base2006, SPECompM2001, SPECompL2001, and Linpack, as well as the compute intensive integer performance benchmark, SPECint_rate_base2006.
Cisco began shipping the Cisco Unified Computing System in July 2009. In less than a year, over 400 customers globally, across all vertical industry sectors, have adopted this innovative architecture because it reduces total cost of ownership – with up to 20 percent reduction in capital expenditures and up to 30% reduction in operational expenditures – increases scalability without adding complexity, and improves IT productivity and business agility.
The Cisco Unified Computing System also helps to create greener data centers by reducing physical footprint, reducing number of components needed, and improving energy efficiency, which significantly reduces power and cooling costs.
"It’s quite an achievement for Cisco to have scored so high on so many of the industry benchmarking tests," said Boyd Davis, General Manager, Data Center Marketing, Intel. "The Cisco Unified Computing System takes full advantage of the Intel Xeon 5600 Series to optimize performance."
"Customers have come to rely on the VMware VMmark benchmark as the standard by which they assess the performance of servers like the Cisco Unified Computing System," said Hatem Naguib, vice president, alliances, VMware. "VMware VMmark concurrently runs multiple ’tiles’ of six virtual machines, including email, Web, database, and file server workloads to test application performance in virtualized environments. The Cisco Unified Computing System demonstrated that it is a system optimized for virtualized environments by achieving a high score that combines virtual machine density and application performance."
"Breakthrough virtualization and application performance benchmarks equate to business impact," said Soni Jiandani, vice president, Server Access and Virtualization Technology Group. "Cisco Unified Computing System innovations such as memory expansion and the Virtualized Network Interface Card, combined with the power of the Intel Xeon 5600 series processor, has created the highest performing 2 socket system available to customers today. These results deliver a clear signal to the market that Cisco’s Unified Computing System is the leading platform for running the most important business applications, and underscores the system’s architectural innovation."
Key Innovations
in the Cisco Unified Computing System Include:
- Memory Extension Technology. With the latest generation Intel processors, virtualized workloads can be bottlenecked by memory rather than CPU. Cisco memory extension technology provides four times the memory (up to 384GB) per server as legacy systems today. The Cisco Unified Computing System B250 M2 Server supports 48 DIMM slots that can be configured to deliver up to 384 GB of main memory using 8-GB DIMMs or up to 192 GB of main memory using lower-cost 4-GB DIMMs. As in the 192GB VMware VMmark benchmark cited above, this delivers unparalleled performance, dramatically reduced system cost, and significant scalability to meet future growth needs.
- Integrated, Cohesive System. The Cisco Unified Computing System provides a cohesive integrated system that unites compute, network, storage access and virtualization into one single system.
- Optimized for a Unified Fabric. Instead of multiple fabrics for network and storage traffic, the system provides one unified fabric to reduce the number of switches, adapters and cables and help simplify management. One-half the support infrastructure is needed compared to legacy systems, reducing capital expenditures, while fewer components to power and cool results in greatly reduced operating expenditures.
- Embedded Management. Management is uniquely integrated into all the components of the system, enabling the entire solution to be managed as a single entity through the Cisco UCS Manager. IT administrators can easily manage 1-320 blades – all as one system. Applications can be provisioned in minutes, compared to days or weeks.
- Virtual Interface Card (VIC). The UCS Virtual Interface Card provides hardware-based support for Cisco VN-Link technology, delivering unmatched performance for virtualized environments using VMware VMDirectPath technology (up to 30% improvement over software based solutions), centralized management through UCS Manager and Service Profiles, and unified fabric support.
- Service Profiles. Service profiles help to automate provisioning and increase business agility, allowing data center managers to provision applications in minutes instead of days. IT managers of storage, networking, compute and applications are able to collaborate on defining service profiles for applications. The service profile stays with the Virtual Machine as it moves, enhancing Virtual Machine portability.
(1)
World record claim based on comparison of two socket server platforms based on x86 architecture. Performance results based on published/submitted/approved results as of March 16, 2010.
(2)
To read all Cisco UCS benchmark
(3)
VMmark configuration details: New result: Published March 16, 2009
The VMmark Benchmark description.
Cisco UCS B250 M2 platform with two Intel Xeon processor X5680 (3.33 GHz, 12MB L3, 6.4 GT/s, 6-core, 130W TDP), Turbo Enabled, HT Enabled, 192GB memory (48x4GB DDR3 1333), Cisco UCS M81KR VIC, EMC CLARiiON CX4-240 storage system with 25x73GB SSD, 20 x 450GB 15K RPM, 5 x 300GB 15K RPM, VMware vSphere 4,0 U1. Score of 35.83@26 tiles.
Published score of 25.16 @ 17 tiles.
SPECint_rate_base2006 Configuration details:
Cisco UCS B200 M2 platform with two Intel Xeon processor X5680 (3.33 GHz, 12MB L3, 6.4 GT/s, 6-core, 130W TDP), Turbo Enabled, HT Enabled, 48 GB (12x4GB DDR3-1333 registered ECC), 1x73GB 15K RPM SAS HDD, SLES 11 Operating system, SPEC binaries were built with Intel Compiler 11.1. Source: Cisco internal measurement as of March 2010. SPECint_rate_base2006 score of 355 and SPECint_rate2006 score of 380. Submitted to www.spec.org for review as of March 16, 2010.
SPECfp_rate_base2006 configuration details:
Cisco UCS B200 M2 platform with two Intel Xeon processor X5680 (3.33 GHz, 12MB L3, 6.4 GT/s, 6-core, 130W TDP), Turbo Enabled, HT Enabled, 48 GB (12x4GB DDR3-1333 registered ECC), 1x73GB 15K RPM SAS HDD, SLES 11 Operating system, SPEC binaries were built with Intel Compiler 11.1. Source: Cisco internal measurement as of March 2010. SPECfp_rate_base2006 score of 248 and SPECfp_rate2006 score of 256. Submitted to www.spec.org for review as of March 16, 2010.
SPECjAppServer2004: configuration details:
Cisco UCS C250 M2 platform with two Intel Xeon processor X5680 (3.33 GHz, 12MB L3, 6.4 GT/s, 6-core, 130W TDP), Turbo Enabled, HT Enabled, H/W Prefetcher Disabled, Adj. Cache Line Prefetch Disabled, DCU Prefecher Disabled, DCU IP Prefetcher Disabled. 96 GB (24x4GB DDR3-1333 registered ECC), 2x73GB 15K RPM SAS HDD, Oracle Enterprise Linux 5 Update 3 x86_64, Oracle WebLogic Server Standard Edition Release 10.3.3, Oracle JRockit 6.0 JDK (R28.0.0-587) (Linux x86 64bit). Database server: Cisco UCS C250 M2 platform with two Intel Xeon processor X5670 (2.93 GHz, 12MB L3, 6.4 GT/s, 6-core), 192 GB (48x4GB DDR3-1333 registered ECC) Oracle 11g R2 database. EMC CLARiiON CX4-240 storage system with 60 x 450GB 15K RPM. Source: Result submitted and approved by www.spec.org as of March 16, 2010. Score of 5,185.45 SPECjAppServer2004 JOPS@Standard.
SPECompMbase2001 configuration details:
Cisco UCS B200 M2 platform with two Intel Xeon processor X5680 (3.33 GHz, 12MB L3, 6.4 GT/s, 6-core, 130W TDP), Turbo Enabled, HT Enabled, 48 GB (12x4GB DDR3-1333 registered ECC), 1x73GB 15K RPM SAS HDD, RHEL 5.4 Operating system, SPEC binaries were built with Intel Compiler 11.1. Source: Cisco internal measurement as of March 2010. SPECompMbase2001 score of 52314 (est). SPECompM results were obtained using 24 OpenMP threads over two sockets. Submitted to www.spec.org for review as of March 16, 2010.
SPECompLbase2001 configuration result:
Cisco UCS B200 M2 platform with two Intel Xeon processor X5680 (3.33 GHz, 12MB L3, 6.4 GT/s, 6-core, 130W TDP), Turbo Enabled, HT Enabled, 48 GB (12x4GB DDR3-1333 registered ECC), 1x73GB 15K RPM SAS HDD, RHEL 5.4 Operating system, SPEC binaries were built with Intel Compiler 11.1. Source: Cisco internal measurement as of March 2010. SPECompLbase2001 score of 278603 (est). SPECompL results were obtained using 24 OpenMP threads over two sockets. Submitted to www.spec.org for review as of March 16, 2010.
Linpack configuration details:
Cisco UCS B200 M2 platform with two Intel Xeon processor X5680 (3.33 GHz, 12MB L3, 6.4 GT/s, 6-core, 130W TDP), Turbo Enabled, HT Disabled, 96 GB (12x8GB DDR3-1333 registered ECC), 1x73GB 15K RPM SAS HDD, RHEL 5.4 Operating system, Intel Linpack 10.2.4. Source: Cisco internal measurement as of March 2010. Linpack score of 146.8 GFlops.