Jason Waxman General Manager High Density Compute Division Data Center Group
Today 2015 More Users Only 25% of the world is Internet connected today 1 New technologies will connect over 1 billion additional users to the cloud 2 More Devices ~80% of Internet connected devices are computers & phones 3 Cars, TVs, households, etc. to increase connected devices 2.5x to >10 billion globally 3 More Content 2.5B photos on Facebook 4 30B videos viewed/mos 5 Google indexes >1T pages 6 8X network, 16X storage & 20x compute capacity needed 7
Economies of Scale Server Consolidation Manageability Low High
FBI seizes servers supporting 350 hosting customers due to criminal investigation of a few customers 12 IT will spend ~$2T on deployment & operations unless smarter infrastructure radically simplifies management of virtualized environments. 13 The cloud consumes 1-2% of the world s energy 11 No specific standards currently exist for enabling interoperability between private clouds or public cloud providers 14
**Estimated savings based on hypothetical optimization case studies. Actual results from such optimizations may vary considerably based on the complexity and numerous variables involved.
\ Acceleration Throughput Performance Efficiency Virtualization Security Power Network & Storage Parallelism Scalability Configurations Manageability Airflow Voltage Regulation Rack Density Cable Management Floor Plan Aisle Layout Integration Operating Conditions
Containers Optimized Racks Dense 1S/2S Servers Micro Servers
1. Power Monitoring: Real-time power consumption Avoid datacenter hotspots Thermal / Power aware scheduling 2. Increase rack density: Enable higher density with power capping 3. Power saving: Workload based power-tuning 4. Power Conservation: Prolong operation during DC outage Workload Pre cap power Pre cap perf Post Cap power Post cap perf Power Cpu intsv Io intsv Memory Workload characterization Mass policy push of lower power state Mix / real Pre outage Post outage
Platform Power Software Datacenter Optimization Management Optimization Efficiency up to up to up to up to $6M $8M $20M $1M With 10% VR efficiency gain 1 With 30W reduced / system 1 With 10% code efficiency upside 1 With 10% PUE improvement 1 1 Values estimated based on a hypothetical 50K server deployment and savings over 3 years to end user, source: Intel, Sept 08 See backup for details
Open Cirrus, UC Berkley RADLab, Universities Intel Cloud Test Bed POCs and Joint Labs Cloud Reference Architectures, Tools and Training Advanced Cloud Research Building Optimized Clouds Deployment Best Practices
Web Transactions per Second 1600 1200 up to 2X 800 400 0 Xeon 5470 Xeon 5570 Visit our booth for a demo! *Source: Parallels, see backup for details
Increased sales by 10% in Six Weeks 30% improvement in Energy Efficiency Increased Application Manageability 60% increase in revenue per server 77% decrease in IT administration cost Up to 100 VPS on Quad core server 30% improvement in Energy Efficiency Reduced Server count, increased new applications through Virtualization
Cloud deployments offer a great economic opportunity to Hosting Service Providers Servers are a critical part of a cloud infrastructure choose a Real Server The Data Center offers many opportunities for optimization Follow well known methods to reduce your TCO and accelerate your service deployments
Accelerate your Cloud Deployments with help from Intel
Legal Disclaimer Intel may make changes to specifications and product descriptions at any time, without notice. Performance tests and ratings are measured using specific computer systems and/or components and reflect the approximate performance of Intel products as measured by those tests. Any difference in system hardware or software design or configuration may affect actual performance. Buyers should consult other sources of information to evaluate the performance of systems or components they are considering purchasing. For more information on performance tests and on the performance of Intel products, visit Intel Performance Benchmark Limitations Intel does not control or audit the design or implementation of third party benchmarks or Web sites referenced in this document. Intel encourages all of its customers to visit the referenced Web sites or others where similar performance benchmarks are reported and confirm whether the referenced benchmarks are accurate and reflect performance of systems available for purchase. Intel processor numbers are not a measure of performance. Processor numbers differentiate features within each processor family, not across different processor families. See www.intel.com/products/processor_number for details. Intel, processors, chipsets, and desktop boards may contain design defects or errors known as errata, which may cause the product to deviate from published specifications. Current characterized errata are available on request. Intel Virtualization Technology requires a computer system with a processor, chipset, BIOS, virtual machine monitor (VMM) and applications enabled for virtualization technology. Functionality, performance or other virtualization technology benefits will vary depending on hardware and software configurations. Virtualization technology-enabled BIOS and VMM applications are currently in development. 64-bit computing on Intel architecture requires a computer system with a processor, chipset, BIOS, operating system, device drivers and applications enabled for Intel 64 architecture. Performance will vary depending on your hardware and software configurations. Consult with your system vendor for more information. Intel, Intel Xeon, Intel Core microarchitecture, and the Intel logo are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries. 2008 Standard Performance Evaluation Corporation (SPEC) logo is reprinted with permission
Xeon 5570-based and Xeon 5470-based Test configurations 3 XEON 5570 web-servers (h2eb1, h2eb2, h2eb3) 2 x Intel Xeon 5570 CPUs, 16 GB RAM each 3 XEON 5470 web-servers (al01, al02, al03) 2 x Intel Xeon 5470 CPUs, 16 GB RAM each 30,000 sites on IPs 10.22.1.100 10.22.1.129 1000 sites per one IP Each site contains several static pages, Wordpress, Joomla and phpbb 1 Load Balancer node (h2eb10) Linux Open source load balancer LVS, included in Linux kernel 2.6.28-rc3 or later 4 MySQL servers for site s databases (h2eb10, h2es4, h2es5, h2es6) 2 x Intel Xeon 5570 CPUs, 16 GB RAM each 4 servers are used as HTTP clients (h2es* and h2eb8) Operating system used: CentOS ver 5.4 (Cloud Linux*)
Backup for TCO Comparison of Various Optimizations** Processor Selection: For IPDCs deploying a 5K server installation, by selecting servers with greater performance and equivalent power consumption, if you are able to reduce footprint by 20% versus plan, the savings could be ~$4M Intel Xeon 5500 versus Intel Xeon 5400 example: 2.5X better performance on SpecWeb 2005, handling far more HTTP requests / second, offering greater work at same energy profile 20% fewer servers: If 5K servers planned for purchase, 20% fewer = 1000 servers avoided CAPEX savings $3M based on $3K / server acquisition cost OPEX savings: Electricity avoidance: 1000 servers at 275W per server running at 60% utilization 3.5 year life span, $.10 / kwhr electricity cost, savings = $500K Cooling avoidance With PUE of 2 = ~$500K cooling cost Total OPEX savings $1 M over 3.5 years Optimizing Power Supplies Optimizing on power supply efficiency for 5K servers: Assume going from 80 percent efficient to 90 percent Assume this is prospective, just like the server performance example A 10% efficiency improvement on a 250W server = 25W Multiplied over 5000 servers Assume $.10 / kwhr Savings in direct power: ~400K Savings in cooling at PUE of 2: ~400K Total savings $800K Improving PUE by 10% Saving 10% PUE in a 5000 server facility Assume the energy consumed is 250W x 5000 servers Assume 65% utilization Assume $.10 / kwhr Energy cost over 3 years = ~$2.5M Cooling cost at PUE of 2 = ~$2.5M Saving 10% of total energy cost = $500k **Estimated savings based on hypothetical optimization case studies. Actual results from such optimizations may vary considerably based on the complexity and numerous variables involved.