Cloud Computing and the i Data Plex Platform

Cloud Computing and the i. Data. Plex Platform Web 2. 0 Expo September 18, 2008 Scott Gerard (sgerard@us. ibm. com) Lab Services, Senior Consultant IBM

Worldwide STG Lab Services Delivery Teams Yorktown Heights, NY Poughkeepsie, NY Mainframe Data Center Services Rochester, MN Speech Technology Hursley, UK 68 Countries Worldwide HPC IBM Training for Systems Dublin, Ireland Data Center Services, Mainframe Power, Modular Data Center Services La. Gaude, France Montpellier, France Mainframe, Power, Modular, Data Center Services West, Central, East Beijing, China IT Consolidation/Virtualization Services (Scorpion) Mainframe, Power, Modular, Storage Mainz, Germany Storage Beaverton, OR Kirkland, WA Modular Tucson, AZ Storage Austin, TX Power RTP, NC A 600 person group delivering Systems implementation, I/T consulting services, and skills development… Modular, Storage Taiwan, Taipei Mainframe, Power, Modular, Storage Bangalore, India Mainframe, Power, Storage Boca Raton, FL Speech Technology Latin America Mainframe, Power, Modular, Storage …part of a team of 25, 000 engineers and programmers in 35 labs in 16 countries 2

IBM New Enterprise Data Center Strategy Leveraging the Best of Traditional and New Practices § Virtualization § Consolidation § Business resiliency and security Traditional Data Centers Web 2. 0 Data Centers § Rapid service delivery § Software resiliency § Pooled shared environment New Enterprise Data Centers • New economics • Rapid service delivery • Aligned with business goals 3

What is driving Cloud Computing • Technology advances that support massive scalability & accessibility • Emergence of data intensive applications & new types of workloads àLarge scale information processing, i. e. parallel computing using Hadoop àWeb 2. 0 rich media interactions àLight weight run anywhere web apps 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. 4

Industry Trends Leading to Cloud Computing A • • • “cloud” is an IT service delivered to users that has: A user interface that makes the infrastructure underlying the service transparent to the user Near-zero incremental management costs when additional IT resources are added A service management platform 2008 2000 1990 Grid Computing • Solving large problems with parallel computing 1998 Utility Computing • Offering computing resources as a metered service • Introduced in late • Made mainstream by 1990 s Globus Alliance Software as a Service • Network-based subscriptions to applications Cloud Computing • Next-Generation Internet computing • Next-Generation Data Centers • Gained momentum in 2001 5

Some Characteristics of Cloud Computing • Virtual – Physical location and underlying infrastructure details are transparent to users • Scalable – Able to break complex workloads into pieces to be served across an incrementally expandable infrastructure • Efficient – Services Oriented Architecture for dynamic provisioning of shared compute resources • Flexible – Can serve a variety of workload types – both consumer and commercial 6

IBM Cloud Computing Gaining Momentum May 2008 April 2008 March 2008 PACES on cloud announced at IMPACT February 2008 VIP/SSME in production on Cloud 2007 Wuxi China Cloud Computing Center Academic Initiative Joint research initiative with 13 European partners Blue Cloud Vietnam Innovation Portal First Cloud Computing Center in Europe Sogeti Online Idea Brainstorm a “terrific success” out of Dublin Cloud Wuxi in production Cloud for i. Data. Plex announced at Web 2. 0 expo Partner to enhance academic research opportunities 7

Worldwide Centers to Serve Clients Dublin, Ireland Seattle, WA Beijing, China San Jose, CA US, East Coast Seoul, S Korea Tokyo, Japan Middle East Bangalore, India Hanoi, Vietnam Singapore São Paulo, Brazil South Africa Announced Planned 8

Dynamic Enterprise Data Center – Enabling Virtual Classrooms Dynamic Scheduling Monitoring Cloud Services Google/IBM Academic Initiative • Promote open standards & Hadoop parallel computing model • Jointly provide compute platform of the future Benefits • Trains students with next generation computing skills • Optimizes emerging Internet scale workloads such as search, video, audio, 3 D Internet, machine learning, mobile computing MIT Carnegie Mellon University of Washington Workloads Virtual Application Server; Virtualization Physical Hardware 9

Academic Initiative University Participants • From Fall 2007 – 2008, over 10 classes taught between 6 universities • Over 500 students trained on next generation parallel computing techniques Projects • Inverted Index • Page. Rank on Wikipedia • Clustering Net. Flix Movie Data • Language Modeling in the Clouds • Large-data Statistical Machine Translation • Collective Resolution of Identity in Email Archives • Parallel Automatic Text-Background Separation in Picture Books • Large-Scale Network Analysis to Improve Retrieval in the Biomedical Domain Students and professors are saying: • “Very cool” • “Job that takes a week now only takes hours” • “Closes the gap between how industry and academia think about computing” 10

Wuxi China Cloud Computing Center IBM establishes the first Cloud Computing Center for software companies in China at the new Wuxi Tai Hu New Town Science and Education Industrial Park in Wuxi, China • Offers emerging Chinese software companies the ability to tap into a virtual computing environment to support their development activities. • A shared facility, providing each company in the Wuxi Software Park with its own virtual data center • Enabled by IBM technology and service • • Managed with IBM Tivoli systems management products Hardware – IBM System x, System p and Blade. Center • Benefits – Fast deployment of Rational software development environments – Up to 200 K software developers, 100 companies – Cost efficient shared infrastructure "The China Cloud Computing Center represents a milestone in service-oriented computing, " said T. W. Liu, the chairman and CEO of i. Soft. Stone. "It will allow companies in the Wuxi Software Park to leapfrog to the newest computing models and will provide an efficient IT platform for software development. " 11

Cloud Computing in the New Enterprise Data Center Workload Solution Patterns Software Development Technology Incubation Innovation Enablement Deploys development tools for immediate use Reduces time to launch new offerings Expands sources of innovation, increases competitiveness Large Scale Information Processing Optimizes emerging Internet scale workloads Cloud Computing Management Services Self-service Admin Portal Workload Pattern Templates Workload Management Administration Workflows Provisioning SLA and Capacity Planning Monitoring Virtualized Physical Servers (Ensembles) i. Data. Plex, Blade. Center, System x, System p, System z 12

Mashups FOR Cloud Management • Data Center much more than just CPUs, storage & networks – Multiple “layers” and multiple types of data/models – Classic system mgmt data doesn’t cover everything – Some models still evolving – Non-standard, client-specific models – Multiple sources: client, vendor, internet, … • Different users have different needs – Needs: user ≠ operator ≠ manager ≠ CIO – Classic system mgmt tools address only a fraction of needs – “Long tail” of needs – Day or less development time. Else too costly 13

Mashups FOR Cloud Management • Integrate multiple data sources – Traditional data center information – Other kinds of information – Social/organizational – Spatial – Thermal/Energy – Vendor product data, documentation/training/background info, … – Internet 14

Enterprise Environment • Draw data from – Intranet – Internet • Keep results inside firewall – Security concerns – Will be difficult to use externally hosted services • Governance – Can enforce compliance with enterprise standards • Minimize risk – Incrementally include more kinds of data – No completely new paradigms 15

Example: Who needs to be notified about server maintenance? Social CI App Components MR Virtual Servers Physical Servers Spatial 16

Layout of Servers in Data Center Annotated with • Server name • Other attributes 17

Mashup Prototype 18

Heat Map Overlaid on Data Center 19

Technologies • Mashup Infrastructure – Lotus Mashup Center – Info. Sphere Mashup Hub – Lotus Widget Factory • Other technologies – Scalable Vector Graphics (SVG, Firefox, Batik) – Semantic Web (RDF, RDFS, SPARQL, OWL) – Integration of “ragged” data – Taxonomies/folksonomies 20

i. Data. Plex • New Technologies: Blade inspired innovation • New Economics: Easy to buy, easy to own • Adaptive Business Model: Aligned End-to-End IBM approach • Business Goal Aligned: Laser vision on achieving our clients’ success 21

Technology Approach Single-Point Management Integrated Rack Deployment Customized Solutions Efficient Power & Cooling Internet Scale Data Center • Custom design layout consulting • Data center-level management software Narrow Depth Rack • Side-by-side chassis • Holistic rack design • Multiple node/chassis combinations Flex Node Technology • Shared power and cooling • High-efficiency power supply • Blade-like technology Green Components • Low-voltage processors and memory • Component elimination • High-efficiency fans 22

Flexible -- Efficient Form and Design 42 U Enterprise Rack 1 42 U Enterprise Rack 2 Top-down view Optional Rear Door Heat Exchanger High Impedance air flow path Typical Enterprise Rack Low impedance air flow path i. Data. Plex Rack 23

New Depth Rack Improves Density No Change to Data Center Layout Sized by floor tiles 400 CFM per tile Cold Isle 2. 4 X Server Density Cold Isle Std Rack w/ 1 U’s Hot Aisle X sq. ft Air Cooling Cold Isle i. Data. Plex Hot Aisle 0. 79 X sq. ft Air Cooling i. Data. Plex Hot Aisle Rear Door Heat e. Xchanger 0. 42 X sq. ft Liquid Cooling 24

A More Intelligent Approach Traditional Rack Servers • 40 servers per rack • 4 network leaf switches • 2 enterprise racks • Configured onsite Internet-Scale i. Data. Plex • 138% better density • 50%+ less floor space • 75% fewer fans • 66% less fan power consumption • $10, 148 energy savings /rack /year • $1. 2 M data center energy savings* • Ships complete, ready to deploy Optional Rear Door Heat e. Xchanger for even greater data center power and cooling efficiency * For typical Data Center 25

Innovative Cooling Solution 15% more Servers 58% less CRACs Only 12% racks are equal to or below 77 F 1 U Air cooled 224 Racks, 9. 4 K Servers (42 / Rack) 100% racks are equal to or below 77 F i. Data. Plex - Rear Door Heat e. Xchanger 128 Racks, 10. 7 K Servers (84 / Rack) 26

IBM Leadership in Dynamic Enterprise Data Centers • Converging Web-centric clouds and enterprise data centers • Establishing worldwide cloud computing centers to drive adoption • IBM leads the way in bringing cloud computing benefits to enterprises 27

Questions 28