Grid Computing Perspectives for IBM Atelier “Internet Et Grilles De Calcul En Afrique”

PSSC – IBM Customer Center Montpellier

Grid Computing Perspectives for IBM Atelier “Internet et Grilles de Calcul en Afrique”

Jean-Pierre Prost IBM France jpprost@fr.ibm.com

Agenda • Grid Computing Initiatives within IBM • World Community Grid • Decrypthon • Cloud Computing and Blue Cloud • Latin American Grid • What about the industry? • Convergence between Grid, Virtualization, and SOA

• IBM has established a Global Public Grid for Philanthropic Research http://www.worldcommunitygrid.org/ • IBM is working with the Mayo Clinic, United Nations, EPA, World Health Organization and United Devices on this worldwide effort

• An advisory Board with members from leading foundations, universities and public organizations is providing oversight to the research projects

• Projects in the following disciplines are being implemented: • Medical Research – Genomics, proteomics, epidemiology, > Over 80 million and biological system research such as AIDS and HIV results returned studies • Environmental Research – Ecology, climatology, pollution, > Over 73,000 years of and preservation CPU processing time • Basic Research – Human health and welfare related studies > Approximately 500k devices on the Grid

AfricaClimate@Home, an ambitious effort based in South Africa Attempt to accurately predict the impact of climate change on various regions in Africa Use the power of 750,000 devices to render the climate models

Join the WCG Community Now ! Employees join World Community Grid at www.worldcommunitygrid.org. Benefits of Employees download a small piece of software (“agent”). Your Company’s Participation Provide employees with an easy, safe, free opportunity to The agent asks World Community make a difference on critical humanitarian issues. Grid for work. Augment employees’ community work at no additional cost. World Community Grid sends work Strengthen corporate citizenship reputation by publicizing to the agent. your company’s overall employee contribution. – Company team will be created, and employee contributions will The agent runs when the PC has earn the team points, as well as individual employee points. unused cycles. – Company listed on “Our Partners” Web page and in all When the agent finishes the materials and public computation, it sends the work back descriptions. to World Community Grid and Join forces with other receives new work. companies, associations, and Company teams and individual public and private employees can see the impact of organizations, working toward a common their contribution and earn points. goal.

Challenge: The Décrypthon program, set up to rapidly accelerate progress in researching genetic diseases and their cures, required an ambitious IT infrastructure to support its mission. Benefits: Solution: – Huge calculation resources now available for IBM, AFM, CNRS and several major French research on genetic diseases and protein universities joined the Décrypthon program to modeling provide a collaborative compute grid including : – Ability to leverage both the unutilised – Building of a nation-wide “University compute power of the universities’ IT Centers” grid leveraging existing HPC infrastructure and also individually-owned infrastructure involving IBM pSeries, and PCs connected to the grid jointly a second “end user” grid to link in the compute resource of the general public – Improved and faster results on genetic disease research meaning that 2 years worth – Implementing specific grid services that are of calculations can now be completed in just mission-critical for this type of project (such 1 year as confidentiality, single-sign-on security and multi-job workload balancing)

“Working with IBM on the implementation of the grid solution for the Décrypthon program will help us optimise the use of the university compute centers and share the compute resources of thousand of end users to get faster results on genetic diseases as part of this ambitious research program” said Stéphane Roques, Secrétaire Général , AFM © 2007 IBM Corporation UniversitiesPSSC – IBM Customer Center Montpellier University Laboratories Computing Centers

IGBMC Orsay Bordeaux Strasbourg Application Server Web Interface IBCP Lyon 16 procesors Pierre et Marie-Curie LORIA Job Nancy Submission and Results Grid Scheduling 8 processors Exploitation MAEM and Management Job Nancy Server Dispatching Lille USTL and Results Université Gathering de Paris 5 16 processors

Orsay Université Data de Paris 6 Storage

Complex Data Management 16 processors Université Requests Lausanne Server Web Interface . CRIHAN - Rouen . Utilisateur de la grille ENS-Lyon Storage © 2007 IBM Corporation PSSC – IBM Customer Center Montpellier Joint IBM Google Announcement

• Train future workforce with next generation IBM Almaden Research computing skills

• University initiative to promote open standards and emerging parallel computing model

• Jointly provide compute platform of the future including hardware, software, and services to support new parallel computing curricula Google

• Three active “clouds”

Universities participating in initial pilot

U. Of Washington

What is Cloud Computing?

An emerging computing paradigm where data and services reside in massively scalable data centers and can be ubiquitously accessed from any connected devices over the internet

Businesses, from startups to enterprises

Web 2.0- 4+ billion phones by enabled PCs, 2010 [Source: Nokia] TVs, etc.

Cloud Computing High-Level Architecture

Apache Virtual Virtual Virtual Virtual Machine Machine Machine Machine

Tivoli Monitoring Agent

Open Source Linux with Xen

Data Center – System x Virtualized Infrastructure based on Open Source Linux & Xen Cloud computing Monitoring and Provisioning infrastructure to support the Management Stack academic initiative can be delivered either as hosted or onsite solution

© 2007 IBM Corporation PSSC – IBM Customer Center Montpellier Blue Cloud Initial Offering – Spring 2008 Delivers a massively scalable and flexible compute platform for hosting both existing and emerging data-intensive workloads Family of ready-to-use cloud computing offerings

Virtualized Infrastructure Based on Linux & Xen

Virtual Virtual Virtual Virtual Machine Machine Machine Machine BladeCenterBladeCenter Tivoli Monitoring Agent

Linux with Xen

Monitoring and Provisioning Cost efficient model for creating and acquiring information Management Stack services Reduces IT management complexity Business Increases business responsiveness with real-time capacity Benefits reallocation as demand for compute power grows Powers both existing and emerging data-intensive workloads

Click above to Start Video Demo

Example of How the Cloud Drives Innovation

Leverage Cloud’s powerful combination of Web 2.0 collaboration platform and dynamic, secure computing environment to drive business transformations

Virtualized Cloud Innovation Factory Incubation Environment

Expand sources of Speed time to market for Lower barriers to IT innovation with a network new offerings by exploiting by leveraging Cloud of partners, customers, collaboration technology to to provide incubation researchers, and academia co-create and gather rapid environment for new feedback on new offerings prototypes

PSSC – IBM Customer Center Montpellier

Education Research

LALA GridGrid

Talent Collaboration Development

Grids are Evolving

UnifiedOpen Offerings Standards Adoption & Interoperability Grid NG Integrated Grids Transactional Grids Virtualization Service Metadata & Orientation Grid 2.0 Global Naming Information Grids Interoperable Components Caching & Data Resource-based Replication Scheduling

Data Movement Scheduling (Grid-FTP) Grid 1.0 Batch & “Pleasingly Parallel” Toolkits & Jobs Custom Integration

Grid & Virtualization

Logical Workload Information Resource Virtualization Virtualization (Distributed Data, Caching, Replication (Dynamic Scheduling) Management Federation, Transformation)

IT System Storage Network Resource Virtualization Virtualization Virtualization (Workload Management, (Virtual Volume (Virtual Devices, Management Partitioning) Management) & Connectivity

Common Resource Model Foundation Service Oriented (SOA) Foundation

SOA Supports Grid How Grid Exploits Service Oriented Architecture

Application components (services) can be easily distributed in as grid Composite Application workload

or moved if resources fail or are overloaded

Enterprise Resources Data Center A Data Center B

Workload can be easily moved to new resources

Open SOA Supports Grid

Foundational web Job description, choreography, workflow, service protocols parallel programming

Scheduling, workload brokering Composite Application

Provisioning, Resource definition, application deployment, endpoint model resource management Enterprise Resources Data Center A Data Center B

• Support the diversity of hardware, operating systems, and software that are common in building grids. • Promote interoperability and allow customers to integrate grid system using middleware from many sources.

Application Layer Orchestration Management Provisioning Workload Information Virtualization Virtualization

Server Network Storage Virtualization Virtualization Virtualization

TPM / TIO XD Workload Scheduler Operating System Loadleveler ITUAM Hypervisor

EWLM Virtual Machine

Virtual Disk

Volume Controller Information Server Network End-point PHYP Grid Scheduler Grid Middleware GPFS NFS TPC SAN Volume Controller PHYP

System z™ IBM IBM System p™ System Storage Servers System i™ System x™

Thank you