Computação Grid IFSC Julho 2005 Sergio Takeo Kofuji

Computação Grid IFSC, Julho 2005 Sergio Takeo Kofuji, Prof. Dr. EPUSP

Motivação Sergio Takeo Kofuji

Three pillars of scientific understanding n n n Theory Experiment Simulation “theoretical experiments” Computational simulation : “a means of scientific discovery that employs a computer system to simulate a physical system according to laws derived from theory and experiment”

Can simulation produce more than “insight”? “The purpose of computing is insight, not numbers. ” — R. W. Hamming (1961) “The computer literally is providing a new window through which we can observe the natural world in exquisite detail. ” — J. S. Langer (1998) “What changed were simulations that showed that the new ITER design will, in fact, be capable of achieving and sustaining burning plasma. ” — R. L. Orbach (2003, in Congressional testimony about why the U. S. is rejoining the International Thermonuclear Energy Reactor (ITER) consortium)

Can simulation lead to scientific discovery? Instantaneous flame front imaged by density of inert marker Instantaneous flame front imaged by fuel concentration Images c/o R. Cheng (left), J. Bell (right), LBNL, and NERSC 2003 SIAM/ACM Prize in CS&E (J. Bell & P. Colella)

The imperative of simulation Applied Physics radiation transport supernovae Environment global climate contaminant transport Biology drug design genomics Engineering crash testing aerodynamics Scientific Lasers & Energy combustion ICF Simulation In these, and many other areas, simulation is an important complement to experiment.

The imperative of simulation Applied Physics radiation transport supernovae Experiments controversial Environment global climate contaminant transport Biology drug design genomics Engineering crash testing aerodynamics Scientific Lasers & Energy combustion ICF Simulation In these, and many other areas, simulation is an important complement to experiment.

The imperative of simulation Experiments dangerous Experiments controversial Applied Physics radiation transport supernovae Environment global climate contaminant transport Biology drug design genomics Engineering crash testing aerodynamics Scientific Lasers & Energy combustion ICF Simulation In these, and many other areas, simulation is an important complement to experiment.

The imperative of simulation Experiments prohibited or impossible Experiments dangerous Experiments controversial Applied Physics radiation transport supernovae Environment global climate contaminant transport Biology drug design genomics Engineering crash testing aerodynamics Scientific Lasers & Energy combustion ICF Simulation In these, and many other areas, simulation is an important complement to experiment.

The imperative of simulation Experiments prohibited or impossible Experiments dangerous Experiments controversial Applied Physics radiation transport supernovae Environment global climate contaminant transport Biology drug design genomics Experiments difficult to instrument Engineering crash testing aerodynamics Scientific Lasers & Energy combustion ICF Simulation In these, and many other areas, simulation is an important complement to experiment.

The imperative of simulation Experiments prohibited or impossible Experiments dangerous Experiments controversial Applied Physics radiation transport supernovae Environment global climate contaminant transport Biology drug design genomics Experiments difficult to instrument Engineering crash testing aerodynamics Experiments expensive Scientific Lasers & Energy combustion ICF Simulation ITER: $5 B In these, and many other areas, simulation is an important complement to experiment.

What would scientists do with 100 -1000 x? Example: predicting future climates n Resolution n New physics n n refine horizontal from 160 to 40 km refine vertical from 105 to 15 km atmospheric chemistry carbon cycle (currently, carbon release is external driver) dynamic terrestrial vegetation (nitrogen and sulfur cycles and land-use and land-cover changes) Improved representation of subgrid processes n n clouds atmospheric radiative transfer

What would we do with 100 -1000 x more? Example: predict future climates Resolution of Kuroshio Current: Simulations at various resolutions have demonstrated that, because equatorial meso-scale eddies have diameters ~10 -200 km, the grid spacing must be < 10 km to adequately resolve the eddy spectrum. This is illustrated in four images of the sea-surface temperature. Figure (a) shows a snapshot from satellite observations, while three other figures are snapshots from simulations at resolutions of (b) 2 , (c) 0. 28 , and (d) 0. 1.

What would scientists do with 100 -1000 x? Example: lattice QCD n n Currently available: 1 Tflop/s Resources at the 100 -200 Tflop/s level will: n n enable precise calculation of electromagnetic form factors characterizing the distribution of charge and current in the nucleon make possible calculation of the quark structure of the nucleon enable calculation of transitions to excited nucleon states Pflop/s resources would: n n enable study of the gluon structure of the nucleon, in addition to its quark structure allow precision calculation of the spectroscopy of strongly interacting particles with unconventional quantum

What would we do with 100 -1000 x more? Example: probe the structure of particles Constraints on the Standard Model parameters r and h. For the Standard Model to be correct, they must be restricted to the region of overlap of the solidly colored bands. The figure on the left shows the constraints as they exist today. The figure on the right shows the constraints as they would exist with no improvement in the experimental errors, but with lattice gauge theory uncertainties reduced to 3%.

Workflow Examples Astronomy Public Health “Collect” Telescope Microscope, Stethoscope, Survey COLLECT “National Virtual Observatory”/ COMPLETE CDC Wonder “Analyze” Study the density structure of a starforming glob of gas Find a link between one factory’s chlorine runoff & disease ANALYZE Study the density structure of all starforming gas in… Study the toxic effects of chlorine runoff in the U. S. “Collaborate” Work with your student COLLABORATE Work with 20 people in 5 countries, in real-time “Respond” Write a paper for a Journal. RESPOND Write a paper, the quantitative results of which are shared globally, digitally.

Workflow a. k. . a. The Scientific Method (in the Age of High. Speed Networks, Fast Processors, Mass Storage, and Miniature Devices) IIC contact: Matt Welsh, FAS

Global. MMCS Web Service MCU Architecture Web Services: Use Multiple Media servers to scale to many codecs and many Session Control audio/video mixing versions of Audio Mixer Video Mixer Session Server Media Servers Web XGSP-based Filters Codec Conversion Control Services Helix Real Streaming PDA Conversion H 323/SIP Gateways Narada. Brokering High Performance (RTP) Thumbnails as NB Scales All Messaging distributed Plus Narada. Brokering Message servers and routers Admire SIP H 323 and XML/SOAP and. . Access Grid Release May 15 Gateways convert to uniform XGSP Messaging As independent can replicate as needed Narada. Brokering Example of stream handling Needs a Grid Farm Native XGSP

Integração de PDA, Cell phone e Desktop Grid Access

GRID n Computação Técnica e Científica Processamento, Dados, Visualisação, Instrumentação n Computadores de Alto Desempenho, Redes de Alta Velocidade n Acesso Restrito n Usuários: Especialistas n n Computação Comercial

Introdução Sergio Takeo Kofuji

Níveis de Grid n Único Domínio Administrativo Departmental n Campus n Corporativo n n Regional Estadual n Nacional n n Global

GRIDs de Aplicações n n n n n Grid Médico & Saúde Biblioteca Digital Multimídia Grid de Sensores Grid de Computação Pervasiva Grid de Física, Biotecnologia, Ambiental Grid de Colaboração Media (Film…) Production and Distribution Grid Rastreamento de Objetos/Animais/Pessoas Jogos

GRID - Tipos Processamento – Supercomputador Virtual Supercomputer; Cluster Virtual etc n Dados/Armazenamento (Storage ) n Intrumentação e Sensor n Visualisação n Colaboração n

GRID – Componentes Físicos n Equipmentos n n n n n Vector/Parallel Supercomputer (NEC SX 7, CRAY) MPPs Clusters (SMP) Servers Desktops Notebooks Palmtops Cell Phones Sensors and Actuators RFIDs

GRID - Tendências Pervasivo n Heterogêneo n Dinâmico n Orientado a Serviços n

Acesso Pervasive ao Grid ‘The Wall’ The GRIDs Environment User Appliance PC Palmtop Mobile. . Plug-in Personal Communication Personal Shopping Hobbies, family activities Business Communication Business Dealing Business Information

Grids - Evolução n 1 a. Geração n n n Computationally intensive, file access/transfer Bag of various heterogeneous protocols & toolkits Recognizes internet, ignores web Academic Team 2 a. Geração n n n Data intensive Knowledge intensive Service based architecture Recognizes Web and Web services Global Grid Forum Industry participation

Grid - Arquitetura Simplificada Typical Application Services Domain Specific Functionality Workflow Algorithms etc Visualization Management Typical Grid Middleware Identification/Scheduling Heterogeneous data management Web Services Technologies Instruments Typical Platform Data Computers Storage Network

Increased functionality, standardization Padrões Abertos de Grid Managed shared virtual systems Computer science research Open Grid Services Arch Web services, etc. Internet standards Custom solutions 1990 Real standards Multiple implementations Globus Toolkit Defacto standard Single implementation 1995 2000 2005 2010

Serviços Grid e Web - Convergência Grid GT 1 Started far apart in apps & tech Web GT 2 OGS I Have been converging HTTP L, WSD WS-* WSRF , SDL 2 W M WSD WSRF - Comunidades de Grid e Web podem se mover para uma base comum

From Savas Parastatidis Orientação a Serviço n Construído sobre os conceitos de Serviço e Mensagens n n A service is the logical manifestation of some physical or logical resources (like databases, programs, devices, humans, etc. ) and/or some application logic that is exposed to the network and A message is a unit of communication for exchanging information. All communication between services is facilitated by the sending and receiving of messages

From Savas Parastatidis Serviço n Contrato Describes the format of the messages exchanged n Defines the message exchange patterns in which a service is prepared to participate n n Política n n Declaratively describe service interaction requirements, quality of service, security, etc Foco em mensagens (messageorientation)

From Savas Parastatidis Troca de Mensagens entre Serviços n Service-orientation (and Web Services) helps architects achieve the following properties (but do not guarantee them) n Scalability, encapsulation, maintenance, re-use, composability, loose coupling, etc.

From Savas Parastatidis Service-orientation vs Resourceorientation Service-orientation Resource-orientation Object-orientation

From Savas Parastatidis Serviço

From Savas Parastatidis Aplicação Orientada a Serviço

From Savas Parastatidis A Cluster-based Service-oriented Application

From Savas Parastatidis An Intranet Service-oriented Application

From Savas Parastatidis An Internet-scale Service-oriented Application

From Savas Parastatidis Serviço Web n Especificações para n n n n Security Orchestration Reliability Policies Federation Management etc.

Peer-to-peer x Grid n P 2 P and Grid share a lot of common ideals n n n P 2 Ps tend to be more dynamic than Grid (Grid resources are usually quite static) P 2 P applications are long-lived (i. e. everyone on the network shares a similar goal of file sharing) n n n Grid applications tend to be transient P 2 Ps often tend to be very fault tolerant n n Both are services communicating by messages on shared resources Multiple redundancy tends to be built in Lack of security is a significant difference between P 2 Ps and Grid P 2 Ps don’t support the idea of VOs effectively (but nothing to stop individuals organizing themselves)

Grid - Níveis Global Grids Multiple enterprises, owners, platforms, domains, file systems, locations, and security policies n Enterprise “Grids” Single enterprise; multiple owners, platforms, domains, file systems, locations, and security policies n SUN SGE EE, Platform Multicluster n Desktop Cycle Aggregation Desktop only United Devices, Entropia, Data Synapse n n Cluster & Departmental “Grids” Single owner, platform, domain, file system and location n SUN SGE, Platform LSF, PBS n Graph borrowed from A. Grimshaw n Legion, Avaki, Globus

OGSA, WSRF, GT 4

Serviço Web - Arquitetura

WS Componentes/Arq n HTTP Server n n Application Server n n Apache AXIS Web Service n n Apache Tomcat SOAP Engine n n Apache HTTP Server You write this Software stack used by GT 4 WSRF Implementation

Serviço Web - Invocação

Serviço Web: Stateless

Serviço Web: Stateful

Access Grid

Access Grids

Access Grid

Access Grid Vision n n To create virtual spaces where distributed people can work together. Challenges: n n Globally Distributed Participants Distributed Resources: n n Computational, Storage, Networks, and People Coordination, Scheduling, Trust Heterogeneous Collaboration Resources Solution: n Deploy a set of Collaboration Resources to serve as the platform for building the rest.

Virtual Venues Client n What can be done: n n Sharing Data Shared Applications Integration with existing Scheduling software n Integrate legacy workflows n APS SER-CAT n n n Applications: n n n Distributed Power. Point Shared Web browser Whiteboard Shared Desktop Tool Shared Visualization Tools n n Fusion Collaboratory n n Beam line Controls Data Processing Data Analysis Archiving/Reporting Collaborative Control Room Atmospheric Science n Shared Data Visualization

The Virtual Venues Client

Wide of Client Platforms Supported Hardware 1. Advanced Node – Tiled Display, Multiple Video Streams, Localized 2. Room Node – Shared Display, Multiple Video Streams, Single Audio 3. Desktop Node – Desktop Monitor, Multiple Video Streams, Single 4. Laptop Node – Laptop Display, Single Video Stream, Single Audio 5. Minimal Node – Compact Display, Single Video Stream, Single Audio Stream Stream Supported Platforms 1. 2. 3. Windows XP/2000 Linux variants (Red. Hat, Slackware, Fedora, Debian, …) Mac OS X (in the future)

Access Grid Security n n Public Key Infrastructure (PKI) based security. Each user, server, and service has an identity cert Communications use SSL (via Globus Toolkit™) SSL provides n Mutual Authentication (each pair of peers knows the identity of the other) Confidentiality Authorization is difficult, so we make it easier n n

Architectural Overview

History and Growth

Crossroads n n Up to now we’ve been serving the AG community (ourselves) exclusively The introduction of this technology can revolutionize science, deploying into new communities will change the social landscape. It’s time for the Access Grid to grow up. So how are we going to get there? n n n Become a service organization? Abandon the software and move on? Adapt and evolve!

Application Deployments n Integration with various user communities n Existing AG Community n n Fusion Collaboration n n > 1000 (@ 40 institutions) ANL Advanced Photon Source CAT Teams n n 1000 public + 300 private: 1000 users 4000 users Center for Learning and Multimodal Communication ABC Collaboration (Surgical Teaching) Introduces new requirements and refinements of existing requirements Increases User base

Toolkit Research Directions n Deeper integration with Grid Computing n n n Investigate much more interesting node solutions Extending Security Work n n n Compute Resources, Data Resources Publish/Subscribe Service Models Peer to Peer Applications and Services More Authentication Solutions More Authorization Solutions Engage new communities

AGTk 2. 2 3. 0 n Network Services n Example Services n n n n Node Management User/Venue Services Chat Improvements Firewall optimizations Port to OS X “Community Service” Operators Interface New Node Services n n Interoperability n n n High Quality Video Display (with layout) Camera Control SOAP (ZSI, Apache) Language (java, C/C++) Grid Data Integration

Roadmap II: AGTk 3. 0 ? ? ? n Advanced Node Configurations n n n Minimal Node Configurations n n n Support for more Active Spaces Better audio environments, high quality video Better media synchronization Integrated instruments (telescopes, beamlines, …) This will cost: Bandwidth Handhelds Set-top box configurations Advanced Collaboration Services n n n Stream processing, modifying Data mining from Streams This will cost: Latency

Roadmap II: AGTk 3. 0 ? ? ? Grid Compute Resources n Authentication Flexibility n Certificate Authority Services n Application Integration (for specific targets) n Workflow Support n Application Hosting Services n

Perguntas? Sergio. kofuji@poli. usp. br www. pad. lsi. usp. br