Dr. Bernard S. Meyerson Cloud Computing Dr. Bernard S. Meyerson IBM Fellow VP Strategic Alliances and CTO IBM Systems & Technology Group
The Enabler; Classical CMOS Scaling Voltage, V /l a WIRING a tox/ W/ a GATE n+ n+ drain source L/a p substrate, dopinga*NA SCALING: Voltage: Oxide: Wire width: Gate width: Diffusion: Substrate: 11Å ~1 nm xd/ a V/a tox /a W/a L/a xd /a a * NA RESULTS: Higher Density: ~a 2 Higher Speed: ~a Power/ckt: ~1/a 2 R. H. Dennard Power Density: ~Constant One Small problem; ATOMS DON’T SCALE! § Approaching atomistic and quantum-mechanical boundaries Why must we follow the recipe at left? • We put 1, 000 times a many devices on a modern chip versus one 40 years ago • The recipe describes how to reduce each device’s power by a factor of 1, 000. Net; Total power is CONSTANT no matter how much we put on a chip • Warning: Failure to follow this recipe may result in setting fire to the user. 2 2 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
The Power Problem; an Industry Dilemma After: R. Schmidt et al. , IBM J. R&D, (2002). ? Opportunity Steam Iron 5 W/cm 2 3 3 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Power Usage in IT – Approaching a Crisis • Datacenter power demands are growing at unsustainable rates – Post “bubble”, spending on datacenter power is growing 600800% faster than spending on servers. – In excess of 80% of Datacenter power is wasted • Source: IDC, Virtualization 2. 0: The Next Phase in Customer Adoption, Doc #204904, Dec 2006 Costs associated with server management and administration have more than tripled over the past decade. 4 4 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Virtualization & Evolution of Data Centers IT Simplification Scale-Out Sprawl Physical Consolidation Windows Servers Firewalls, Routers Unix Servers V Management Servers Virtual Servers, Storage, Networks Ensemble V Networks V V Linux Servers Service Oriented Data Center, Ensembles, & Cloud Services V Mainframe or Unix Server Switches Abstraction and Pooling Storage Linux Server Storage Multi-System Virtualization V Servers Ensemble V V Networks Ensemble Storage Driving forces for Data Center Advancement: • Sufficient network connectivity • Enhanced disaster recovery and simplified administration • Reduced labor and real estate costs • Improved server price/performance ratio 5 5 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Forces Driving Cloud Computing • My. Space took 25 months to reach 20 M users • You. Tube took 16 months to reach 20 M users Skyrocketing costs of power, space, maintenance, etc. Explosion of data intensive applications on the Internet Advances in multi-core computer architecture Fast growth of connected mobile devices Growth of Web 2. 0 enabled PCs, TVs, etc. 6 6 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Building the cloud 7 7 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Holistic Design: A “New” Paradigm in Value Creation Innovation from Atoms to Services • The simultaneous optimization of the following: following Materials, Devices, Circuits, Cores, Chips, System Architecture and Assets, Software and Services • Provides the most effective means to optimize the value of IT offerings to the end user Note: Execution relies upon the seamless integration of skills from across the spectrum 8 8 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
IBM Innovations across the stack Technology Systems Software • SOI • Copper • Immersion Technology • Multicore • 3 -D Chip Stacking 9 9 • System Z • Websphere • i. Data. Plex • Rational • System P • Tivoli Data Center • Portable Data Center • Water Cooling • Mobile Measurement Technology • Monitoring & Provisioning Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Internet Scale Data Centers Increase Flexibility in Managing Power • Monitoring, Visualization, & Integration with Tivoli portfolio • Power-aware Placement Controller –Placement for energy efficiency –Consolidation of workloads Node 1 A B C • Autonomic Workload-aware Management of Energy Throttles CPU Clock to Save Power Placement Controller A B Node 2 A B Node 3 C 10 10 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Energy Management Policy Example APP 4 APP 3 APP 2 APP 1 Workload Migration Enables Dynamic Server Consolidation Use of hibernation, powering off servers, and other low power states in combination with other workload balancing and provisioning tools can provide a valuable tool in management of Power and Thermal issues. System 2 APP 6 APP 5 System 1 System 3 Automate Energy Control Policy-based automation Control Energy Consumption Consolidate workloads to reduce 11 11 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
The “Smart” Datacenter Integrate Real-time IT, Facilities, and Utility Infrastructure Monitoring and Control Utility Facility Alternative UPS Network Power Battery IT Security… Utility Cooling Tower Pumps Chiller te pu Utility, Rates, Incentives $ Substation Communicating Revenue Meter IT and Networks Ice H 2 O SOA Convergence Utility Generator Cooling m Co ge ra o St k or etw , N r we Po ing l w ro oo In- d C an Precision Cooli ng Central UPS Power Distribution Units Parallel or Transfer Eqpt Medium Voltage >600 VAC Eqpt CHP Fuel Cell, Micro. Turbine or Turbine Power Low Voltage 600 VAC Eqpt Raised Floor DC Power Compute • Main Frames • Volume Servers • Blade Servers Storage • SATA Disk • Tape • Blended Network • Corporate Networks • Vo. IP • Integrated Blade/Switch • Network closets In-Row Power • Modular UPS • Rack Mount PDUs In Row Cooling • Rear Door Heat Exchanger • Liquid Cooling Racks • Overhead Cooling Service Level Agreements 12 12 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Enterprise-Wide Real-time Infrastructure Data Center A Tier III, Lots of Capacity Low electricity €, £ or $ IT and Facility Operation Center IT and Facilities viewing same dashboards with real-time info Optimized Systems SOA, ITIL reusable processes Data Center D Tier II, No Capacity Stranded capacity Data Center B Tier II, Marginal Capacity High electricity €, £ or $ Data Center C Cyclone or hurricanes in area 13 13 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Blue Cloud 14 14 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
The IBM Blue Cloud™ Initiative “Deliver Cloud Computing and IT Simplification to our customers, integrating the best of IBM's existing and future products to simplify the deployment and management of customer workloads. ” Windows Servers Virtual Servers, Windows Server Storage, Networks Firewalls, Unix Servers Routers Unix Server Switches Networks Pools of Virtualized Resources Management Servers Storage Linux Servers Storage Service Networks Storage Complex Physical Consolidation Virtualization Ensemble Cloud 15 15 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Enterprise Cloud Computing Evolution to the New Enterprise Data Center A New Enterprise Data Center Web-centric Cloud e. g. Google System z, System x, System p, Blade. Center Enterprise Data Center • Request driven, dynamic, virtualized, scalable • Shared infrastructure for multiple workloads • Optimized for security, transactions, data integrity • Dynamic information delivery • Accessed from anywhere • Shared, optimized for traffic, scalable • Mission critical transactions • Controlled access • Optimized for security and integrity 16 16 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Cloud Computing in the New Enterprise Data Center Workload Solution Patterns Software Development Technology Incubation Deploys development tools for immediate use Reduces time to launch new offerings Large Scale Information Processing Innovation Enablement Expands sources of innovation, increases competitiveness Optimizes emerging Internet scale workloads Cloud Computing Management Services Self-service Admin Portal Virtualized Physical Servers Workload Management Provisioning & Monitoring SLA and Capacity Planning System Z, System X, System P, Bladecenter 17 17 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
IBM delivers Cloud today Technology Incubation Blue Cloud Computing* Collaborative Innovation Government-led Initiatives Data Intensive Workloads Academic Initiative Software Development *In 2007, IBM announced its Blue Cloud initiative. “Blue Cloud” is not a branded offering but rather an IBM Internal designation used to describe the collection of enabling technologies that IBM uses to create the Cloud Computing experience for its customers in the data center, service provider and Cloud hosted environments. 18 18 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
IBM Cloud Computing Centers serve clients around the world Dublin, Ireland Seattle, WA Beijing, China San Jose, CA US, East Coast Seoul, S Korea Tokyo, Japan Middle East Wuxi, China Bangalore, India Hanoi, Vietnam São Paulo, Brazil Active Planned South Africa Source: Hi. PODS 19 19 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Wuxi China Cloud Computing Center Announcement • Announced the first commercial Cloud Computing Center, February 1 st 2008 • A shared facility, providing each company in the Wuxi Software Park with its own virtual data center • Managed with IBM Tivoli systems management products • Hardware – IBM System x, System p and Blade. Center • Software - IBM Rational, Web. Sphere, DB 2 database 20 20 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Wuxi China - Cloud Computing for Software Development • Software companies share a secured virtualized facility, isolated from each other, to accelerate software outsourcing tasks • Service oriented data center (SOA) • Automated software installation • Instantaneous development environments to individual developers Company A Developer Virtual Machine Company C Administrator Specifies computing needs for “on demand” fulfillment Virtual Machine Web 2. 0 Admin Interface Virtual Machine Virtual Machine Company B Tester Virtual Machine Virtual Machine Tivoli Monitoring Agent Open Source Linux with Xen System x System p Blade. Center 21 21 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Vietnam Ministry of Science and Technology leverages cloud to run its innovation program VISTA Innovation Portal (VIP) VIP pilot hosted on IBM’s Blue Cloud computing infrastructure at Almaden Students Teachers Blogs Profiles Wikis Forums Social Tagging Information Discovery Researchers IBM Innovation Factory VIP, powered by IBM Innovation Factory, provides a platform to foster collaborative innovation among major universities and research institutes. 22 22 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Utilizing the cloud with disruptive technologies Stream as a Service? 23 23 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
The explosion of events requires extreme innovation Traditional Computing Historical fact finding with data-at-rest Stream Computing Real time analysis of data-in-motion Streaming data § Batch paradigm, pull model § Query-driven: submits queries to static data § Relies on Databases, Data Warehouses • A stream of structured or unstructured data-in-motion Stream Computing • Analytic operations on streaming data in real-time 24 24 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Stream processing will be everywhere…. Stock market • Impact of weather on securities prices Natural Systems • Seismic monitoring system • Water management • Precision agriculture Air Transportation Law Enforcement • Real-time multimodal surveillance Fraud prevention • Global air traffic management • Detecting multi-party fraud • Real time fraud prevention Manufacturing • Process control for microchip fabrication Health & Life Sciences Radio Astronomy • Detection of transient events • Epidemic early warning system • Remote healthcare monitoring 25 25 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
If lifespan Stream of data usefulness is Computing measured in seconds automated Operational Business Intelligence Event Processing Platform Stream computing analyzes events at extreme speeds response is required If action time can be several hours, Traditional then traditional Systems days or more, systems can handle it Action time: from micro-second to minutes, hours Graph source: Syndera Action time: duration between time when relevant event happens and time when corresponding action is taken 26 26 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Example: System S Processing Building Blocks Transform NYSE Dynamic P/E Ratio Calculation Millions of Events VWAP Calculation SEC Edgar 10 Q Earnings Extraction Caption Extraction Video News Weather Data Speech Recognition Hurricane Weather Data Extraction Filter Annotate Correlate Join P/E with Aggregate Impact Earnings Moving Average Calculation Topic Filtration Earnings Related News Analysis Hurricane Forecast Hurricane Model 1 Forecast Hurricane Model 2 Forecast Hurricane Model … Forecast Model N Earnings News Join Hurricane Risk Encoder Trade Decision Hurricane Impact Hurricane Industry Impact Classify < 1 millisecond latency 27 27 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Scaling streams using cloud computing from a single instance to thousands VWAP Timeperiod NYSE Calculate P/E Ratio Deployment on Cluster High Performance the Cloud SEC Edgar News Weather Data RSS Parser Keyword Filter Join P/E and Aggregate Impact Semi Annual Moving Average Keyword Filter Gister RSS Parser Keyword Filter Hurricane Forecast Hurricane Impact Hurricane Industry Impact VWAP Timeperiod NYSE Calculate P/E Ratio SEC Edgar News Weather Data RSS Parser Keyword Filter Join P/E and Aggregate Impact Semi Annual Moving Average Keyword Filter Gister RSS Parser Keyword Filter Hurricane Forecast Trade Decision Hurricane Impact Hurricane Industry Impact VWAP Timeperiod NYSE Calculate P/E Ratio SEC Edgar News Weather Data RSS Parser Keyword Filter Semi Annual Moving Average Keyword Filter Gister RSS Parser Keyword Filter Hurricane Forecast Join P/E and Aggregate Impact Hurricane Industry Impact 28 28 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
IBM’s Vision for Cloud Computing • Holistic approach to Cloud Computing integrates all technical and business disciplines to attack challenges through innovation across many different fields. • Cloud computing will provide fast time to market and refocus IT on customer application value and less on IT management. • The flexibility and ease-of-use of cloud computing will enable experimentation, leading to better results, and wins in the marketplace. 29 29 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Backup 30 30 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Priority Spending on Virtualization Server virtualization Server consolidation SOURCE: Goldman Sachs Group, Inc, July 2008 % of respondents rating issue as high priority 31 31 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Re-convergence to a Centralized Infrastructure Source: Merrill Lynch, The Cloud Wars: $100+ billion at stake, May 2008 Cloud computing enjoys the best of both worlds: distributed personal computing of the 90's and centralized mainframe computing of the 70 s 32 32 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
The Cloud over the years Network-Delivered Services are the culmination of a long term trend to simplify the purchase of IT Services 1961: John Mc. Carthy proposes computing as a utility 1961: IBM Services Bureau 1975: First inter-industry EDI standards 1981: SMTP defines the standard electronic mail service 1985: United Nations sponsors EDIFACT 1990: Berners-Lee invents the World-Wide Web 1994: Commerce. Net 1998: Rosetta. Net 1999: i-Mode mobile internet 2000: IBM BCRS 2000: UDDI 1. 0; “Saa. S” coined 2001: Dot com bubble bursts IBM Service Bureau (1961) 2005: IBM Ao. D 2006: Amazon EC 2 2007: Google Health; force. com launch 2008: IBM ww Cloud Computing centers 19601970 1980 1990 2000 2010 33 33 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Tivoli Monitors and Manages Energy across the Data Center Integrated Asset and Facilities Management Green Service Management Integrating IT and Power Facility Design, Maintenance and Remediation Business Service Management Energy Usage and Accounting Green Management Data Warehouse IT Server Provisioning and Load Balancing IT Assets (Servers, storage) Energy cost modeling Data Warehouse IT Assets Energy Manager Data Center Infrastructure (UPS, PDUs) IT Asset Lifecycle Management Application Response Time Management Additional ITM Agents Building Systems (HVAC, power, lighting, security) 34 34 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
IBM’s Cloud Offerings Cloud Consulting & Implementation Services Public Cloud Services Strategic Outsourcing Cloud Services Examples: • Evaluate a Cloud’s architecture against the IBM Cloud Architecture Standards for Resiliency, Availability, Continuity/Recoverability, Scalability, and Security. • Help customers create their own Enterprise Cloud infrastructure as part of transforming their data centers. • Use economic models for assessing the TCO for building private clouds, and/or moving data and applications off-site in a public or hybrid cloud model to help clients choose the cloud solutions 35 35 Dr. Bernard S. Meyerson, Parallels Summit, Feb. 2009© Copyright IBM Corporation 2008
Скачать презентацию Dr Bernard S Meyerson Cloud Computing Dr Bernard
31c6b836b5149c14b954322f6ee3d56d.ppt