In this week’s episode, Robert Novak (@gallifreyan) and Steve McQuerry (@smcquerry) talk about Big Data on UCS. Real world use cases including UCS Manager, UCS Central, and more.
Big Data, Big Unicorn, courtesy of Robert Novak and Steve McQuerry
This is Engineers Unplugged, where technologists talk to each other the way they know best, with a whiteboard. The rules are simple:
- Episodes will publish weekly (or as close to it as we can manage)
- Subscribe to the podcast here: engineersunplugged.com
- Follow the #engineersunplugged conversation on Twitter
- Submit ideas for episodes or volunteer to appear by Tweeting to @CommsNinja
- Practice drawing unicorns
Join the behind the scenes by liking Engineers Unplugged on Facebook.
Tags: Big Data, Cisco UCS, Hadoop, UCS Central, UCS Manager
When we talk about Cisco UCS success, we could boast about how, in just five short years, Cisco has earned the #1 position for x86 blade server sales in the Americas. Or, brag on how we achieved 39% revenue growth in a period where the other top five vendors saw flat or declining numbers, according to IDC. We could mention how some in the industry questioned our strategy to enter this market in the first place, in the middle of a worldwide economic slowdown no less. Also, we might remind everyone that others predicted that we wouldn’t last a year.
We could do that, but we won’t. Because we know without a doubt, that the success of Cisco UCS is not all about us. It is about our customers, who continue to choose innovation. Like our customers, we are acutely aware that the data center, including the role of IT, is changing rapidly. Cost pressures, technology changes, and game-changers like cloud computing require a new approach. With our unified data center architecture, products, solutions, and services, we are helping you achieve the business outcomes you need for your organization to grow. We are working with you, over 33,000 strong and growing, to help evolve your data centers using the Cisco Unified Data Center platform and redefine the economics of your IT operations. So you can spend more of your resources on delivering innovation to your business, all while increasing business agility, reducing CapEx and OpEx, and simplifying your operations.
It’s also about Cisco data center partners, because we know that it takes an ecosystem to be successful. Driving 80% of all Cisco UCS sales, channel partners are fueling this rapid growth and capturing emerging opportunities including cloud, big data, enterprise applications, and desktop virtualization. Early on, our partners recognized the value of UCS and embraced it. Their training, infrastructure, and other investments have been a huge contributor to UCS traction in the marketplace. The results are an expanded data center practice and value delivered to customers. Cisco data center partners, we couldn’t have done it without you.
Not long ago, Paul Perez predicted this day would come. His bold “sooner than you think” statement has come to pass, well, sooner than we thought. Frankly, we’re not too surprised. But we are pleased, and humbled, and determined to keep pushing forward with data center innovation such as Cisco Application Centric Infrastructure.
But most of all, we thank you.
Tags: Cisco UCS, data center, x86 blade
What is integrated infrastructure? Major analysts firms differ slightly on the taxonomy and makeup of the various integrated systems that comprise this market segment. However, they do agree that these systems include a combination of servers, networking, storage systems, and management. Cisco realized that bringing together industry and market leading technologies would help accelerate IT infrastructure deployment, lower design costs, simplify management, and enable high levels of utilization. Integrated infrastructures offer the foundation for private cloud deployments, virtual desktop solutions, and enterprise applications.
Everyone agrees that the integrated infrastructure market is booming and that Cisco is a core vendor in this market segment with leading integrated infrastructures built on top of Cisco UCS and Cisco Nexus. IDC believes this market is going to grow at greater than 30% for the next few years to more than $14B in 2017. Today, Cisco UCS and Cisco Nexus participate in the top three Integrated Infrastructures –FlexPod Solutions with NetApp, Vblock Systems with VCE, and EMC VSPEX. FlexPod and Vblock have captured more than 42% of the market and utilize Cisco UCS and Nexus exclusively.
In addition, Hitachi Unified Compute Platform Select and Nimble Storage SmartStack also chose to build integrated infrastructure solutions with Cisco UCS servers and Cisco Nexus switches. If you add up the numbers from IDC, Cisco participates in integrated infrastructures that account for roughly 69% of the market from a revenue perspective. Very impressive!
It is also impressive that just five years ago we announced Cisco UCS and it is now the #1 provider of x86 Blade Servers in the Americas and #2 World Wide (according to the most recent IDC Worldwide Quarterly Server Tracker – May 2014). Cisco UCS has delivered some ground breaking technology. The policy-based Service Profiles and Cisco SingleConnect technology changed how we connected, managed, and provisioned servers. Cisco continues to drive this strategy forward with Application Centric Infrastructure (ACI), offering comprehensive control of a much broader set of data center infrastructure that is oriented to the application environment. FlexPod, Vblock Systems, and Cisco solutions for EMC VSPEX plan to integrate Cisco ACI with Cisco UCS Director to accelerate application delivery and streamline operations. Both NetApp and VCE recently announced they will also integrate Cisco’s Intercloud Fabric with FlexPod and Vblock Systems, enabling the management of cloud-enabled workloads across heterogeneous environments.
This rapid rate of customer adoption for Cisco UCS and integrated infrastructures enabled by Cisco is evidence that real innovation can deliver IT infrastructure that transforms the data center. For more information on integrated infrastructures enabled by Cisco please check out this new white paper.
Tags: Cisco UCS, Converged Infrastructure, FlexPod, Integrated infrastructure, Vblock, vspex, x86 blade servers
Cisco highlighted its support for OpenStack at the recent OpenStack Summit in Atlanta, which hosted 4500+ attendees and included many more users, in addition to the developers and operators that have dominated past conferences. A common theme among keynote presentations was the speed and flexibility of IT required to support the clouds that will soon dominate commerce and communication worldwide. The effort underway to improve stability was also a recurring discussion topic.
OpenStack Summit, May 12-15 in Atlanta
From its beginning as an open source project at NASA, the OpenStack movement has grown as an open alternative to propriety cloud services and applications. The Summit serves as a forum for those interested in hashing out the direction and adoption of the model and standards, as well as a learning opportunity for those ready to build and deploy on them.
Keynote speakers from Wells Fargo and Disney helped transition the Summit from an academic exercise to a forum for learning how innovative companies are taking control of their cloud environments.
Glenn Ferguson, Head of Private Cloud Enablement for Wells Fargo, described the compliance, auditing and governance Wells requires in its private cloud, that aren’t available in public cloud offerings. Wells has designated OpenStack their “cloud infrastructure model” to facilitate rapid deployment of infrastructure to meet application developers’ needs and requires all IT vendors to work within the OpenStack specifications. “This is something we have to do to remain agile and competitive in this environment,” Ferguson said. “Our infrastructure needs to keep pace with the software.”
Chris Launey, Disney’s Director of Cloud Architectures and Services, was blunt in how he described the value of speed. “If you’re a business that deals in any kind of information, you need speed (to thrive.) “If you give (developers) their own ‘fast’, they’ll make their own ‘cheap’ by getting their product to market quickly and responding to customer demands. And (they’ll) make their own ‘good’ by shrinking development cycles and introducing improvements more often, until they reach a virtual continuous cycle of improvements.”
The OpenStack Foundation divides the work into individual projects focused on the various cloud components: servers, object-based storage, networking infrastructure, security, etc. Proponents are excited about the innovation that can be unleashed when developers are freed from having to worry about the complexities associated with underlying infrastructure and can focus on the innovation of cloud services and applications.
Cisco was highly visible at the Summit, drawing standing-room-only crowds to sessions in the Networking Track, as network stability and scalability are top-of-mind for users deploying critical applications and services to an open source cloud.
Lew Tucker, Cisco Vice President and CTO for Cloud Computing and Vice-Chair of the OpenStack Foundation, painted a picture of what is possible in his presentation “Open Stack and the Transformation of the Data Center.” He described how the data center is becoming a large, highly automated “fabric” consisting of interconnected physical systems and virtualized services. In this environment, OpenStack acts as a platform for building a highly efficient cloud, providing management of diverse infrastructure “below” and orchestration of a vast set of application services “above”.
Lew Tucker, Cisco VP and CTO of Cloud Computing
Cisco’s key contribution to OpenStack has been participation in the development of Neutron, the OpenStack Networking Service. There is clearly a need to have the same level of visibility and management flexibility that Cisco has been offering its customers in an open source cloud model. In addition to driving connectivity generally, Cisco has received approval on blueprints for plugins to integrate VPN- and Firewall-as-a-Service as part of OpenStack networking. (Referred to as Network Function Virtualization (NFV) plugins.) Cisco is also working on the integration of OpenStack Neutron with OpenDaylight, a separate project started to focus specifically on network programmability. Cisco’s extensive work in the open source community will bring even greater value to its existing customers by extending the ecosystem of solutions integrated with Cisco products.
In the Expo Hall, Cisco highlighted the integration of its networking, compute and management products with OpenStack APIs, demonstrating:
If you missed the Summit, check out the Session Videos and Slides to deep-dive presentations by Cisco contributors, presented at the Atlanta Summit 2014:
Tags: ACI, Cisco, Cisco UCS, cloud, devops, Neutron, nexus, NFV, opendaylight, OpenStack
Cisco UCS continues its tradition of performance leadership with the announcement of three world –record benchmark performance results:
- Cisco UCS B260 M4 blade servers achieved the best 2-processor 2-tier SAP Sales and Distribution (SD) Benchmark performance
- Cisco UCS C460 M4 Rack Server set new world-record SPECjbb®2013 MultiJVM 4-socket max-jOPS performance result.
- Cisco UCS B260 M4 Blade Server claimed a new world-record SPECjbb®2013 MultiJVM 2-socket x86/64 max-jOPS performance result
Let’s take a closer look at each of these world-record benchmark performance results:
1. World-Record: Fastest Two-Processor Result for SAP
Cisco UCS B260 M4 blade servers delivered the best 2-processor SAP Sales and Distribution (SD) Benchmark result, in a two-tier configuration, with performance accelerated by Cisco UCS Invicta™ Series Solid State Systems. The Cisco UCS B260 M4 Blade Server running Microsoft Windows Server 2012 delivered a world-record two-tier SAP SD Benchmark result on SAP Enhancement Package 5 for SAP ERP 6.0 and Microsoft SQL Server 2012. The solution supported 12,280 SAP SD Benchmark users while maintaining a consistent application response time of less than one second. Published results can be found on the SAP website at http://global.sap.com/solutions/benchmark/sd2tier.epx, SAP SD Benchmark Result Certificate Number 2014018.
Cisco UCS B260 M4 delivered a SAPS score of 67,020. This result is an 18 percent improvement over the 54,700 SAPS score delivered by IBM x3690 X5 servers configured with previous-generation processors and a 22 percent improvement over an IBM FlexSystem p260 running two IBM POWER 7+ processors as reflected in the graph below:
Check out the Performance Brief for additional information on the benchmark configuration. The detailed official benchmark disclosure report is available at the SAP website.
2. The Same Processors: Greater SPECjbb2013 Performance with the Cisco UCS C460 M4
Cisco UCS C460 M4 Rack Server captured the top 4-socket MultiJVM score for maximum Java operations (max-jOPS) with results of 201,117 max-jOPS and 52,784 critical-jOPS.
The test configuration consisted of a controller and four groups, each consisting of a transaction injector and back end, all running across nine JVM instances within a single operating system image. The JVM instances ran on a Cisco UCS C460 M4 Rack Server powered by four 2.8-GHz, 15-core Intel Xeon processor E7-4890 v2 CPUs running the Red Hat Enterprise Linux (RHEL) 6.5 operating system and 64-bit Oracle Java HotSpot Virtual Machine (VM) server on Linux Version 1.8.0.
The benchmark results of 201,117 max-jOPS and 52,784 critical-jOPS place the Cisco UCS C460 M4 at the top of the max-jOPS scores for 4-socket servers running multiple JVMs and outperforms the Dell PowerEdge R920 and HP ProLiant DL580 Gen8 servers, as shown in the graph below:
Check out the Performance Brief for additional information on the benchmark configuration. The detailed official benchmark disclosure report is available at spec.org website.
3. Cisco Improves Performance by Up to 47 Percent in World-Record SPECjbb2013 Results
Cisco UCS B260 M4 Blade Server captured the top 2-socket x86/64 MultiJVM score for maximum Java operations (max-jOPS) with results of 91,499 concurrent Java operations per second and 30,021 concurrent critical Java operations per second on SPECjbb2013.
The Cisco configuration consisted of a controller and two groups, each consisting of a transaction injector and back end, all running across multiple JVM instances within a single operating system image. The JVM instances ran on a Cisco UCS B260 M4 Blade Server powered by two 2.8-GHz, 15-core Intel Xeon processor E7-4890 v2 CPUs running the Red Hat Enterprise Linux (RHEL) 6.4 operating system and Oracle Java HotSpot 64-Bit Server Virtual Machine (VM) on Linux Version 1.8.0.
The benchmark result of 91,499 max-jOPS and 30,021 critical-jOPS places the Cisco UCS B260 M4 at the top of the max-jOPS scores for 2-socket x86/64 servers running multiple JVMs delivering a 47 percent increase in SPECjbb®2013 max-jOPS and a 28 percent increase in critical-jOPS compared to Cisco’s recent results with the Intel Xeon processor E5 v2 powered servers as depicted in the graph below:
Check out the Performance Brief for additional information on the benchmark configuration. The detailed official benchmark disclosure report is available at spec.org website.
These results join the more than 90 previous world-record performance results set by Cisco UCS since it was announced five years ago. Many of these world-record results use real-world workloads, including business applications, Java middleware, database performance, high-performance computing, and virtualized and cloud computing environments. Check out the Cisco UCS Performance Benchmarks website for additional details.
For additional information on Cisco UCS and Cisco UCS solutions please visit Cisco Unified Computing & Servers web page.
Competitive claims based on results posted at http://www.spec.org/, and at http://www.cisco.com/c/en/us/products/servers-unified-computing/industry_benchmarks.html as of 05/29/2014. SPEC and SPECjbb are trademarks or registered trademarks of Standard Performance Evaluation Corporation.
SAP Results referenced are from the SAP website at http://global.sap.com/solutions/benchmark/sd2tier.epx and are current as of 5/29/2014: Cisco UCS B260 M4, 2 Processors / 30 Cores / 60 Threads, Intel Xeon Processor E7-4890 v2, 2.8 Ghz, 64 KB L1 cache and 256 KB L2 cache per core, 37.5 MB L3 cache per processor, 512 GB of main memory, Microsoft Windows Server 2012 Datacenter Edition, Microsoft SQL Server 2012, and SAP Enhancement Package 5 for SAP ERP 6.0; certification number 2014018, IBM Flex System p260 Compute Node, 2 processors, 16 cores, 64 threads, IBM POWER7+, 4.10 GHz, 32-KB (I) and 32-KB (D) L1 cache and 256-KB L2 cache per core, 10- MB L3 cache per core, 256 GB of main memory, IBM AIX 7.1, IBM DB2 10, and SAP Enhancement Package 5 for SAP ERP 6.0; certification number 2012035, IBM System x3690 X5, 2 processors, 20 cores, 40 threads, Intel Xeon processor E7-2870, 2.40 GHz, 64- KB L1 cache and 256-KB L2 cache per core, 30-MB L3 cache per processor, 256 GB of main memory, Microsoft Windows Server 2008 Enterprise Edition, IBM DB2 9.7, and SAP Enhancement Package 4 for SAP ERP 6.0; certification number 2011032
Senior Marketing Manager,
Unified Computing System
Tags: Cisco UCS, enterprise applications, Performance Benchmarks, Perofrmance Leadership, SAP SD, SPEjbb2013