Grid Computing and The Gridbus Toolkit: Creating and Managing Utility Grids for e. Science and e. Business Applications Dr. Rajkumar Buyya Fellow of Grid Computing and Distributed Systems (GRIDS) Lab. Dept. of Computer Science and Software Engineering The University of Melbourne, Australia gridbus. org/~raj/tut/gridbus. zip WW Grid
4 Essential Utilities (in Home) (1) Water (2) Electricity (3) Gas (4) Telephone 2
(5) IT services as the fifth utility (water, electricity, gas, telephone, IT) 3 e. Science e. Business e. Government e. Health Multilingual e. Education …
GRIDS Lab @ Melbourne R&D n The youngest and one of the largest research labs in the CSSE Dept: n n n n 4 Academics Industries Software: n n Faculties of Science, Engineering, and Medicine Many national and international collaborations. n n National and International organizations Australian Research Council Many industries (Sun, Storage. Tek, Microsoft, IBM) University-wide collaboration: n n 2 Post. Docs 2 Research Programmers 7 RHD (6 Ph. D) students ~5 honours/masters projects Funding n n Education Our Grid middleware technologies are widely in academic and industrial users. Publication: n My research team produces 20% of our Dept’s research output. + Community Services
Books at Glance: Coauthored/edited 5
Presentation Outline n Part 1: Introduction to Grid Computing and Applications n n n Technology Evolution and Application Drivers Grid Challenges, Approaches, and Architecture Part 2: Grid Economy and Service Oriented Computing n Challenges Service-Oriented Grid Architecture (SOGA) Realisation of SOGA n Grid Market Directory, Grid. Bank, VPM, Grid Service Broker, G-Monitor n n Part 3: Global Grids and Gridbus Technologies Part 4: Performance Evaluation on the World-Wide Grid n n n Part 5: Closing Remarks n n n 6 Compute Grid Application e. Science Application – Belle Analysis Data Grid Open Challenges in Grid Economy Analogy to Electric Power Grid Summary and Conclusion
Evolution: Humans e. Humans (e. Hugging, e. Smell, e. Food!), Science e. Science, Business e. Business 7
COMPUTING Computing and Communication Technologies Evolution * HTC * Mainframes * Minicomputers * PCs * Workstations * P 2 P * Grids * Computing Utility * PC Clusters * Crays * XEROX PARC worm * PDAs * MPPs * WS Clusters Communication * e. Science * TCP/IP * Sputnik 1960 8 * Ethernet * Email 1970 1975 * W 3 C * HTML * Mosaic * Internet Era * ARPANET * e. Business * IETF 1980 1985 * WWW Era 1990 * Web Services * XML 1995 2000 * Social. Net 2010
Computing Evolving towards: Global/Grid Computing S E R V I C E S 2100 2100 2100 + Administrative Barriers P E R F O R M A N C E 9 • Individual • Group • Department • Campus • State • National • Globe • Inter Planet • Universe Personal Device SMPs or Super. Computers Local Cluster Enterprise Cluster/Grid Global Grid Inter Planet Grid
Leading to Grid (computing) Paradigm: Cyberinfrastructure for sharing resources • Inspired by Power Grid! • * A service-oriented/utility computing paradigm that enables seamless sharing of geographically distributed, autonomous resources. • * 10 This was the original aim of building Internet although it ended up in giving birth to email!
What is Grid ? (there are several definitions) n A type of parallel and distributed system that enables the sharing, selection, & aggregation of geographically distributed “autonomous” resources: n n Wide area n n n Computers – PCs, workstations, clusters, supercomputers, laptops, notebooks, mobile devices, PDA, etc; Software – e. g. , ASPs renting expensive special purpose applications on demand; Catalogued data and databases – e. g. transparent access to human genome database; Special devices/instruments – e. g. , radio telescope – SETI@Home searching for life in galaxy. People/collaborators. depending on their availability, capability, cost, and user Qo. S requirements. 11
A Bird Eye View of World-Wide Grid Environment Grid Information Service Grid Resource Broker R 2 R 3 R 5 Application database R 4 RN Grid Resource Broker R 6 Grid Information Service 12 R 1 Resource Broker
Type of Services Modern Grids Offer n n n 13 n Computational Services – CPU cycles n NASA IPG, WWG, Tera. Grid, SETI@Home Data Services n Data replication, management, secure access--LHC Grid/Napster Application Services n Access to remote software/libraries and license management—Net. Solve Information Services n Extraction and presentation of data with meaning Knowledge Services n The way knowledge is acquired and managed using meta data & semantics. Utility Computing Services Utility Grid Knowledge Grid Information Grid ASP Grid Data Grid Computional Grid
Prominent Grid Drivers: Emerging e-Science and e-Business Apps n n Next generation experiments, simulations, sensors, satellites, even people and businesses are creating a flood of data. e-Science refers to the large scale science that will increasingly be carried out through distributed global collaborations enabled by the Internet. ~PBytes/sec High Energy Physics Brain Activity Analysis Newswire & data mining: Natural language engineering Life Sciences Digital Biology Quantum Chemistry 14 Astronomy Finance: Portfolio analysis Internet & Ecommerce
1. Online Medical Instrumentation and Neuroscience Osaka Univ. Hospital DV transfer Analysis Results Data Generation Virtual Laboratory for. Data Analysis brain science medicine and Analysis • Knowledge sharing Results Cybermedia Center • MEG sharing? • Data Sharing 15 Life-electronics laborat AIST • Provision of MEG • Provision of expertise in the analysis of brain fun
3. Enterprise Computing Applications n Traditional Model n Grid Based Model Service Virtualization Layer & Load Balancing Email server 16 Web server Database server Apps server Upgrade to a new server to handle more users Horizontal integration of Email, Web, Data, and Apps servers
Global Grids and Challenges
E-Science / E-Business App Elements E-Scientist Peers sharing ideas and collaborative interpretation of data/results Distributed instruments Distributed computation 2100 Remote Visualization Distributed data 18 Data & Compute Service 2100
Grids have Emerged as Cyberinfrastructure that scales from enterprise to global Grid Information Service Grid Resource Broker R 2 R 3 R 5 Application database R 4 RN Grid Resource Broker R 6 Grid Information Service 19 R 1 Resource Broker
Grid-based Utility Computing model need to scale from desktops to Global level S E R V I C E S 2100 2100 2100 + Administrative Barriers P E R F O R M A N C E 20 • Individual • Group • Department • Campus • State • National • Globe • Inter Planet • Universe Personal Device SMPs or Super. Computers Local Cluster Enterprise Cluster/Grid Global Grid Inter Planet Grid
Grids need to offer a wide variety of services n n n 21 Computational Services – CPU cycles n SETI@Home, NASA IPG, Tera. Grid, IGrid, … Data Services n Data replication, management, secure access--LHC Grid/Napster Application Services n Access to remote software/libraries and license management—Net. Solve Information Services n Extraction and presentation of data with meaning Knowledge Services n The way knowledge is acquired and managed—data mining. Utility Computing Services n Towards a market-based Grid computing: Utility Grid Knowledge Grid Information Grid ASP Grid Data Grid Computional Grid
Grid Challenges Security Computational Economy Uniform Access Resource Discovery 22 Resource Allocation & Scheduling Application Construction System Management Data locality Network Management
Grid Operations Management Challenges – dynamic resources, policies, and self interested entities GSP G O C GSP Grid Exchange GSP 5 GSP 1 Grid Economy Technologies GSP 2 23 GSP 4
Some Grid Initiatives Worldwide n USA n n Australia n n n Brazil n n n China 120 million – 5 yrs n n Europe n 450 million – 5 yrsn 486 million – 5 yrsn n India 1. 3 billion (Rs) § n Japan n Nimrod-G Gridbus DISCWorld Grange. Net. 27 million APACGrid ARC e. Research n n n Our. Grid, Easy. Grid LNCC-Grid + many others n n UK e. Science EU Grids. . and many more. . . I-Grid NAGERI Singapore NGP n n n IBM On Demand Computing HP Adaptive Computing Sun N 1 Microsoft -. NET Oracle 10 g Infosys – Enterprise Grid Satyam – Enterprise Grid Storage. Tek –Grid. . and many more Public Forums n n n Global Grid Forum Australian Grid Forum Conferences: n n 24 1. 3 billion – 3 yrs Industry Initiatives n China. Grid – Education CNGrid - application 1 billion – 5 nyrs Korea. . . N*Grid n n Globus Grid. Sec Access. Grid Tera. Grid Cyberinfrasture and many more. . . CCGrid HPDC E-Science http: //www. gridcomputing. com 2? billion
mix-and-match (service) Object-oriented Internet/partial-P 2 P Network enabled Solvers Economic-based Utility / Service-Oriented Computing Nimrod-G 25
The Gridbus Project @ Melbourne: Enable Leasing of ICT Services on Demand Distributed Data WWG Gridbus World Wide Grid! On Demand Utility Computing 26
27
Presentation Outline n Part 1: Introduction to Grid Computing and Applications n n n Technology Evolution and Application Drivers Grid Challenges, Approaches, and Architecture Part 2: Grid Economy and Service Oriented Computing n Challenges Service-Oriented Grid Architecture (SOGA) Realisation of SOGA n Grid Market Directory, Grid. Bank, VPM, Grid Service Broker, G-Monitor n n Part 3: Global Grids and Gridbus Technologies Part 4: Performance Evaluation on the World-Wide Grid n n n Part 5: Closing Remarks n n n 28 Compute Grid Application e. Science Application – Belle Analysis Data Grid Open Challenges in Grid Economy Analogy to Electric Power Grid Summary and Conclusion
Gridbus considers: “Incentive” as a Design Parameter for Grid Computing n Grids aim at exploiting synergies that result from cooperation of autonomous distributed entities. Synergies include: n n n 29 Creation of Virtual Organisations/Enterprises Resource sharing Aggregation of resources on demand. For this cooperation to be sustainable, participants needs to have (economic) incentive. Therefore, “incentive” mechanisms should be considered as one of key design parameters of Grid computing.
Grid Economy: Methodology for Sustained Resourced Sharing and Managing Supply-and-Demand for Resources 30
Benefits of Computational Economy n n It provides an effective paradigm for managing self interested and self-regulating entities (resource owners and consumers) Helps in regulating supply-and-demand of resources. n n n User-centric / Utility driven Scalable: n n n n 31 Services can be priced in such a way that equilibrium is maintained. No need of central coordinator (during negotiation) Resources(sellers) and also Users(buyers) can make their own decisions and try to maximize utility and profit. Adaptable, It allows to offer different Qo. S (quality of services) to different applications depending the value users place on them. It offers incentive for resource owners for being part of the grid! It offers incentive for resource consumers for being good citizens. It improves the utilisation of resources.
It helps Users to Achieve their Goals n Grid Consumers n n n Execute jobs for solving varying problem size and complexity Benefit by selecting and aggregating resources wisely Tradeoff timeframe and cost n n Grid Providers n n n Contribute (“idle”) resource for executing consumer jobs Benefit by maximizing resource utilisation Tradeoff local requirements & market opportunity n 32 Strategy: minimise expenses Strategy: maximise return on investment
New challenges of Grid Economy n n n 33 Grid Service Providers (GSPs) n How do I decide service pricing models ? n How do I specify them ? n How do I translate them into resource allocations ? n How do I enforce them ? n How do I advertise & attract consumers ? n How do I do accounting and handle payments? n …. . Grid Service Consumers (GSCs) n How do I decide expenses ? n How do I express Qo. S requirements ? n How do I trade between timeframe & cost ? n How do I map jobs to resources to meet my Qo. S needs? n …. . They need mechanisms and technologies for value expression, value translation, and value enforcement.
GRACE: A Reference Grid Economy Services Architecture GRid Architecture for Computational Economy (GRACE)
Market-based Computing Systems Requirements n To enable users (GSPs and GSCs) to realise economic value, market-based systems need to provide mechanisms for: n Value Expression n n Value Translation n n scheduling policies to translate them to resource allocations Value Enforcement n n 35 a means to express their requirements, valuations, and objectives mechanisms to enforce the selection and allocation of differential services, and dynamic adaptation to changes in their availability at runtime Market mechanisms, accounting and payment, Reservation of resources.
GRACE: A Reference Service-Oriented Grid Architecture for Computational Economies Data Catalogue Grid Bank Job Control Agent Grid Node N Secure Schedule Advisor Qo. S Grid Node 1 Grid Resource Broker Trading … Deployment Agent Job. Exec Misc. services Resource Allocation Storage Grid Middleware Services Accounting Resource Reservation R 1 Grid Consumer 36 Pricing Algorithms Trade Server Trade Manager Health Monitor … Grid Explorer Info ? Information Service … Programming Environments Applications Sign-on Grid Market Services R 2 … Rm Grid Service Providers
Realising Market-based Grid: Minimal New Components n n Grid Market Directory Services Grid Trading Services – n n 37 for different economic models Grid Metering Services Grid Accounting and Payment Services Grid Service Broker
Gridbus and Complementary Grid Technologies – realizing GRACE Science Commerce … MPI Excell. Grid Brokers: Grid Economy Alchemi Nordu. Grid Windows Gridscape Unicore … XGrid JVM Solaris Collaboratories Workflow Engine Nimrod-G Globus . NET Engineering Grid Storage Economy Condor Linux … IRIX Libra Tomcat Mac OSF 1 PDB Worldwide Grid Core Grid Middleware Grid Market Directory CDB 38 User-Level Middleware (Grid Tools) Gridbus Data Broker SGE AIX X-Parameter Sweep Lang. Grid Exchange & Federation Grid Bank PBS … Grid Applications Portals G R I D S I M Grid Fabric Software Grid Fabric Hardware
On Demand Assembly of Services: Interaction Between Grid Components Application Code Explore data 1 Data Source Visual Application Composer Data + 10 ults fo Res t In Cos 2 Data Catalogue 6 Grid Resource Broker ASP Catalogue lts 9 8 Grid Service (GS) (Globus) 4 Grid Info Service 3 Job Data Replicator (GDMP) 5 Grid Market Directory 7 Alchemi Bill GS Cluster Scheduler PE 39 Grid Service Provider (GSP) (e. g. , CERN) CPU or PE GSP (e. g. , IBM) 12 Resu (Instruments/dis tributed sources) Cluster Scheduler PE GSP (e. g. , Uof. M) PE GTS GSP (e. g. , VPAC) 11 Gridbus Grid. Bank GSP (Accounting Service)
40
41
Alchemi: . NET-based Enterprise Grid Platform & Web Services Alchemi Manager Web Services Internet Alchemi Users Internet 42 • SETI@Home like Model • General Purpose • Dedicated/Non-dedicate workers • Role-based Security • . NET and Web Services • C# Implementation • Grid. Thread and Job Model Programming • Easy to setup and use • Widely in use! Alchemi Worker Agents
Some Users of Alchemi Tier Technologies, USA Large scale document processing using Alchemi framework Satyam Computers Applied Research Laboratory, India Micro-array data processing using Alchemi framework CSIRO, Australia Natural Resource Modeling The University of Sao Paulo, Brazil The Alchemi Executor as a Windows Service The Friedrich Miescher Institute (FMI) for Biomedical Research, Switzerland Patterns of transcription factors in mammalian genes 43 stochastix Gmb. H, Germany Asynchronous Excel Tasks using Managed. XLL and Alchemi. Net Grid Computing framework. Many users in Universities: See next for an example.
Students' project gives old computers new life 1/25/2005 44
d 45
Globus Technologies Usage n Security (GSI - Globus Security Infrastructure) - single sign-on and authentication based on RSA public key cryptography technology. n n Information (MDS - Metacomputing Directory Service) – LDAP-server based uniform access to resource structure/state information. n n n 46 You need have Grid ID, public key, and private key (assigned by trusted CA) Authorization to use: You need have your Grid ID mapped to a physical (login) account on every Grid nodes that you want to use. Authentication: User proxy (trigger by grid-proxy-init) and Grid node gatekeeper authenticate each other by exchanging messages. (If you can decrypt the message that I sent by encrypting using your public key, then you are who you are claiming to be. ) GIIS – Grid Index Information Service (one for your Grid!/organisation) GRIS – Grid Resource Information Service (one for each node). Communications (grid-ftp) - multi-method communication and Qo. S management. Process/Job Management (GRAM - Globus Resource Allocation Manager) Low-level (uniform) API for various local schedulers. Remote file access (GASS - Global Access to Secondary Storage). Reservation of Resources in Advance (GARA).
Globus Components (in One Slide) Client-side APIs MDS client API calls to locate resources MDS: Grid Index Info Server Site boundary MDS client API calls to get resource info GRAM client API calls to MDS: request resource allocation and process creation. GRAM client API state change callbacks Globus Security Grid Resource Info Server Query current status of resource Local Resource Manager Infrastructure Request Create Gatekeeper Job Manager Parse RSL Library 47 Monitor & control Allocate & create processes Process
Presentation Outline n Part 1: Introduction to Grid Computing and Applications n n n Technology Evolution and Application Drivers Grid Challenges, Approaches, and Architecture Part 2: Grid Economy and Service Oriented Computing n Challenges Service-Oriented Grid Architecture (SOGA) Realisation of SOGA n Grid Market Directory, Grid. Bank, VPM, Grid Service Broker, G-Monitor n n Part 3: Global Grids and Gridbus Technologies Part 4: Performance Evaluation on the World-Wide Grid n n n Part 5: Closing Remarks n n n 48 Compute Grid Application e. Science Application – Belle Analysis Data Grid Open Challenges in Grid Economy Analogy to Electric Power Grid Summary and Conclusion
49
The Grid Market Directory Grid Vision: To enable the creation of Virtual Enterprise (VE), Virtual Oranisation (VO), or Grid Market. Place (GMP).
A Market-Oriented Grid Environment 51
Grid Market Infrastructure n Grids need to provide an infrastructure that supports: n n 52 (a) the creation of one or more GMP registries; (b) the contributors to register themselves as GSPs along with their resources/application services that they wish to provide; (c) GSPs to publish themselves in one or more GMPs along with service prices; and (d) Grid resource brokers to discover resources/services and their attributes (e. g. , access price and usage constraints) that meet user Qo. S requirements.
GMD Architecture Grid Market Directory (GMD) Web Server (Tomcat) GMD Portal Manager GMD Query Webservice Grid Service Info (RDBMS) Publish/Manage Browse Query(SOAP+XML) Job submission Grid Node Provider (Web Client) Grid Node 53 Consumer (Grid Resource Broker) Consumer (Web Client)
Globus MDS Vs Gridbus GMD Globus MDS 54 Gridbus GMD
GSP Registration 55
GSP Service Publication 56
GSP Service Browsing 57
GMD Query Message GMD Repository XML SOAP Message HTTP Server Query Message GMD Webservice client SOAP Engine Repository Handler Query Message Query Processing GMD Query Webservice 58
GMD Use Case: SC’ 02 HPC Challenge Demonstration 59
How can I Access GMD Software ? n Download, Deploy, and Use it: n n Or Make use of Global GMD registry hosted by the Gridbus Project. For more info, Read Technical Report: n 60 “Open Source” Reference Implementation (Java-based) is available: http: //www. gridbus. org/gmd/ A Market-Oriented Grid Directory Service for Publication and Discovery of Grid Service Providers and their Services
61
Grid Bank A Grid Accounting Services Architecture
The Grid Bank Operations Grid. Bank Server Grid. Cheque + Resource Usage (GSC Account Charge Grid. Cheque User A p p l i c a t i o n s Establish Service Costs Grid Resource Broker (GRB) Grid. Bank Payment Module Grid Trade Server Grid. Bank Charging Module Grid. Cheque Resource Usage Grid Agent Deploy Agent and Submt Jobs Grid Resource Meter Usage Agreement R 1 R 2 R 3 Grid Service Consumer (GSC) User Grid Service Provider (GSP) 63 R 4
Grid. Bank Architecture 64
Grid Bank Components n Grid Bank Server n n n Grid. Bank Database Grid. Bank Client Access Interface n n 65 Regular account management features (open, close, delete, update, browse) are supported. Payment Module Charging Module Protocols in XML format Resource Usage Record (confirm to GGF RUR format).
User Applications Grid Components Interaction and Utilization of Grid Bank Grid Resource Broker (GRB) Grid. Cheque Grid. Bank Payment Module Grid. Bank system component names are in italics Grid. Bank Server Charge Establish Service Rates Grid Trade Server Execute job Grid. Cheque Grid Resource Meter + Resource Usage Record Grid. Bank Charging Module RATES Item 1 – Rate Item 2 – Rate. . . 66 Convert to standard Resource Usage Record RUR X X X Usage – Item 1 Usage – Item 2. . . = Charge for Item 1 = Charge for Item 2. =. =. = Service Cost Total Filter relevant resource usage information R 1 R 2 R 3 R 4 Grid Service Provider (GSP) Chargeable Item 1 – Rate Chargeable Item 2 – Rate. . .
Grid Bank Usage Scenario n n n 67 n GSPs and GSCs open account with Grid. Bank When GSC wants to consume GSP service, it informs the GSP about the account to which access cost can be charged. GSPs can confirm with Grid. Bank whether GSC has sufficient credit or even request to put the amount on hold. GSP measures the amount of resource consumed and charges the GSC account in Grid Bank transfers to tokens/credits/money from GSC to GSP account; and maintains transaction details (Resource Usage Record). Grid Bank also be used for developing Scalable Authentication Infrastructure.
Access Scalability Problem Resource access authorization file (grid-mapfile) “/O=Grid/O=Globus/OU=cs. uwa. edu. au/CN=Alexander Barmouta” alex “/O=Grid/O=Globus/OU=cs. mu. oz. au/CN=Rajkumar Buyya” rajkumar Clients “/O=Grid/O=Globus/OU=cs. uwa. edu. au/CN=Chris Mc. Donald” chris X 509 v 3 Digital Certificate ………… Subject: “/O=Grid/O=Globus/OU=cs. uwa. edu. au/CN=Alexander Barmouta” ………… X 509 v 3 Digital Certificate ………… Subject: “/O=Grid/O=Globus/OU=cs. mu. oz. au/CN=Rajkumar Buyya” ………… 68 X 509 v 3 Digital Certificate ………… Subject: “/O=Grid/O=Globus/OU=cs. uwa. edu. au/CN=Chris Mc. Donald” ………… Resources
Grid. Bank’s Solution to Access Scalability Problem Grid. Bank Accounts “/O=Grid/O=Globus/OU=cs. uwa. edu. au/CN=Alexander Barmouta” “/O=Grid/O=Globus/OU=cs. mu. oz. au/CN=Rajkumar Buyya” Request to access resource “/O=Grid/O=Globus/OU=cs. uwa. edu. au/CN=Chris Mc. Donald” Passing client’s Certificate Subject Execute job Template (local) accounts gbaccount 1 gbaccount 2 gbaccount 3 Resource access authorization file (grid-mapfile) “/O=Grid/O=Globus/OU=cs. uwa. edu. au/CN=Grid. Bank” gridbank “/O=Grid/O=Globus/OU=cs. uwa. edu. au/CN=Alexander Barmouta” gbaccount 1 69
How can I Access Grid. Bank Software ? n Download, Deploy, and Use it: n n n For more info, Read Technical Report: n 70 “Open Source” Reference Implementation is available: http: //www. gridbus. org/ Grid. Bank: A Grid Accounting Services Architecture (GASA) for Distributed Systems Sharing and Integration
71
The Gridbus Grid Service Broker for Data Grid Applications Builds on the Nimrod-G Computational Grid Broker and Computational Economy [Buyya, Abramson, Giddy, Monash University, 1999 -2001] And Extends its notion for Data and Service Grids
Grid Service Broker (GSB) n n n A resource broker for scheduling task farming data Grid applications with static or dynamic parameter sweeps on global Grids. It uses computational economy paradigm for optimal selection of computational and data services depending on their quality, cost, and availability, and users’ Qo. S requirements (deadline, budget, & T/C optimisation) Key Features n n n n 73 n A single window to manage & control experiment Programmable Task Farming Engine Resource Discovery and Resource Trading Optimal Data Source Discovery Scheduling & Predications Generic Dispatcher & Grid Agents Transportation of data & sharing of results Accounting
Gridbus Broker Architecture Gridbus Client App, T, $, Opt (Bag of Tasks Applications) (Data Grid Scheduler) Gridbus Farming Engine Schedule Advisor Trading Manager Record Keeper Grid Dispatcher Grid Explorer Grid Middleware TM $ TS GE GIS, NWS Grid Info Server RM & TS $ $ U G Globus enabled node. 74 G L Data Node C Unicore enabled node. A RM: Local Resource Manager, TS: Trade Server Alchemi enabled node. Data Catalog
Gridbus Broker and Remote Service Access Enablers Portlets Home Node/Portal Gridbus Broker Credential Repository My. Proxy batch() fork() -PBS -Condor -SGE -Alchemi -XGrid Data Catalog Alchemi Globus Data Store Job manager Unicore Access Technology fork() batch() -PBS -Condor -SGE 75 Gridbus agent Grid FTP SRB Gateway SSH fork() batch() -PBS -Condor -SGE -XGridbus agent
Gridbus Services for e. Science applications n Application Development Environment: n n n Resource Allocation and Scheduling n n Dynamic discovery of optional computational and data nodes that meet user Qo. S requirements. Hide Low-Level Grid Middleware interfaces n 76 XML-based language for composition of task farming (legacy) applications as parameter sweep applications. Task Farming APIs for new applications. Web APIs (e. g. , Portlets) for Grid portal development. Threads-based Programming Interface Workflow interface and Gridbus-enabled workflow engine. Globus (v 2, v 4), SRB, Alchemi, Unicore, and ssh-based access to local/remote resources managed by XGrid, Condor, SGE.
Click Here for Demo Figure 3 : Logging into the portal. 77 Drug Design Made Easy!
Excel. Grid Middleware Excel. Grid Add-In Excel. Grid Runner Excel. Grid. Job Excel Plugin to Access Gridbus Services 78 Gridbus Broker Enterprise Grid 210 0 210 0
Adaptive Scheduling Steps Discover Establish Resources Rates Distribute Jobs 79 Compose & Schedule Discover More Resources Evaluate & Reschedule Meet requirements ? Remaining Jobs, Deadline, & Budget ?
Deadline (D) and Budget (B) Constrained Scheduling Algorithms Algorithm Execution Compute Time (D) Cost (B) Grid Cost Opt Limited by D Minimize Yes Cost-Time Opt Minimize if possible Minimize Limited by B Yes Conservative -Time Opt Minimize Limited by B, jobs have guaranteed minimum budget Yes Yes Time Opt Data Grid 80 Yes
81
82
Case Study: High Energy Physics and Data Grid n The Belle Experiment n n 83 KEK B-Factory, Japan Investigating fundamental violation of symmetry in nature (Charge Parity) which may help explain the universal matter – antimatter imbalance. Collaboration 400 people, 50 institutes 100’s TB data currently
Australian Belle Data Grid Platform 84
Case Study: Event Simulation and Analysis B 0 ->D*+D*-Ks • Simulation and Analysis Package - Belle Analysis Software Framework (BASF) • Experiment in 2 parts – Generation of Simulated Data and Analysis of the distributed data Analyzed 100 data files (30 MB each) were distributed among the five nodes 85
Resources Used and their Service Price Organization Role Cost (in G$/CPUsec) CS, Uni. Melb belle. cs. mu. oz. au 4 CPU, 2 GB RAM, 40 GB HD, Linux Broker host, Data host, NWS server N. A. (Not used as a compute resource) Physics, Uni. Melb fleagle. ph. unimelb. edu. au 1 CPU, 512 MB RAM, 40 GB HD, Linux Replica Catalog host, Data host, Compute resource, NWS sensor 2 CS, University of Adelaide belle. cs. adelaide. edu. au 4 CPU (only 1 available) , 2 GB RAM, 40 GB HD, Linux Data host, NWS sensor N. A. (Not used as a compute resource) ANU, Canberra belle. anu. edu. au 4 CPU, 2 GB RAM, 40 GB HD, Linux Data host, Compute 4 resource, NWS sensor Dept of Physics, USyd belle. physics. usyd. edu. au 4 CPU (only 1 available), 2 GB RAM, 40 GB HD, Linux Data host, Compute 4 resource, NWS sensor VPAC, Melbourne 86 Node details brecca-2. vpac. org 180 node cluster (only head node used), Linux Compute resource, NWS sensor 6
Network Cost (in Grid $/Currency!) 87
Deploying Application Scenario n n A data grid scenario with 100 jobs and each accessing remote data of ~30 MB Deadline: 3 hrs. Budget: G$ 60 K Scheduling Optimisation Scenario: n n n 88 Minimise Time Minimise Cost Results:
Time Minimization in Data Grids fleagle. ph. unimelb. edu. au belle. anu. edu. au belle. physics. usyd. edu. au brecca-2. vpac. org 80 70 Number of jobs completed 60 50 40 30 20 10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 Time (in mins. ) 89
Results : Cost Minimization in Data Grids fleagle. ph. unimelb. edu. au belle. anu. edu. au belle. physics. usyd. edu. au brecca-2. vpac. org 100 90 80 Number of jobs completed 70 60 50 40 30 20 10 0 1 90 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 Time(in mins. )
Observation Organization Node details Cost (in G$/CPUsec) Total Jobs Executed Time Cost CS, Uni. Melb N. A. (Not used as a compute resource) -- -- Physics, Uni. Melb fleagle. ph. unimelb. edu. au 1 CPU, 512 MB RAM, 40 GB HD, Linux 2 3 94 CS, University of Adelaide belle. cs. adelaide. edu. au 4 CPU (only 1 available) , 2 GB RAM, 40 GB HD, Linux N. A. (Not used as a compute resource) -- -- ANU, Canberra belle. anu. edu. au 4 CPU, 2 GB RAM, 40 GB HD, Linux 4 2 2 Dept of Physics, USyd belle. physics. usyd. edu. au 4 CPU (only 1 available), 2 GB RAM, 40 GB HD, Linux 4 72 2 VPAC, Melbourne 91 belle. cs. mu. oz. au 4 CPU, 2 GB RAM, 40 GB HD, Linux brecca-2. vpac. org 180 node cluster (only head node used), Linux 6 23 2
The Grid. Sim Toolkit A Java based tool for Grid Scheduling Simulations Application, User, Grid Scenario’s Input and Results Application Configuration Resource Configuration Visual Modeler Grid Scenario . . . Output Grid Resource Brokers or Schedulers’s Simulation Grid. Sim Toolkit Application Modeling Resource Entities Information Services Job Management Resource Allocation Statistics Resource Modeling and Simulation (with Time and Space shared schedulers) Single CPU SMPs Clusters Load Pattern Network Reservation Basic Discrete Event Simulation Infrastructure Sim. Java Distributed Sim. Java Virtual Machine (Java, c. JVM, RMI) PCs 92 Workstations SMPs Clusters Distributed Resources Add your own policy for resource allocation
Selected Grid. Sim Users - 2002 93
Workflow Scheduling SKIP (if Time Problem)
Grids and Workflow Issues: • Naming • Security store GET dataset galfit store • Authorization • Service interface • Programming models • Work flow • etc. sextr stack • Data representation and interchange GET dataset phot preview store Astronomical Data Analysis (Hugh Couchman, Computing in Canadian Astronomy) 95
Grid-based workflow n Grid workflow n n Differences n n n 96 A collection of tasks that are processed on distributed resources in well-defined order. Grid workflow could be long lasting Large data flow need to be supported (e. g. Sloan Digital Sky Survey ~Petabytes) Resources used by Grid workflow are heterogeneous Resources are dispersed across multiple administrative domains Resource availability and utilization varies dynamically over time
Requirements and Challenges n Requirements n n Challenges n n n 97 Composition tools (e. g. expressing large-scale workflow) Harnessing distributed resources and services that meet user requirements Large-scale data transfer Dynamic execution environment of Grid workflow Unknown locations of intermediate data Acquisition of resource information
Workflow Management System n n 98 Developed a service-oriented workflow management system driven by IBM TSpaces Provides XML based language for expressing workflow Able to deploy workflow applications on global grids Serve as an infrastructure for our future work on economy-based workflow scheduling.
Architecture Workflow Planner Application Composition …… Scientific Portal Workflow description & Qo. S Workflow Submission Handler Workflow Enactment Engine Info Service Workflow Language Parser GMD Resource Discovery Tasks Parameters Dependencies Replica Catalog Workflow Scheduler Dispatcher Gridbus Broker Globus MDS Data Movement Web services HTTP Grid. FTP r sfe t an a tr Da Database 99
Workflow Scheduling System n n Workflow Coordinator (WCO) n TM generation and activation n Life-time of workflow execution Task Managers (TMs) n Task execution n Resource discovery and selection n Monitoring n n Failure management Task Manager Factory Task Manager Event Service Monitor Resource Group Task Group Decentralized Scheduling Architecture Communication approach between WCO and TMs n Communication Model n n n Complexity of task dependencies (e. g. multiple parents and multiple children) Many-to-many Solutions n n n 100 Workflow Coordinator Event-driven mechanism Subscription-notification Event exchange server using tuple spaces (IBM TSpaces)
Event-driven Mechanism using Tuple Spaces Workflow Coordinator notify Event Service (IBM TSpaces) notify Monitor status output notify Task Manager A Grid resources 101 Task Manager B . . . Task Manager N
A Sample WF model, Task and Datalink Definition A Fa Fa B Fb E Fa C Fc Fe Fb G Fg D Fd Fc F Ff H Directed Acyclic Graph 102 Fd <task name="C"> optional <executable> <name>ycalc</name> <host>belle. anu. edu. au</host> <accesspoint type="GT 2 Gram">/data/ycalc. sh</accesspoint> <input> <port 0 type="file">para</port 0> <port 1 type="msg">5</port 1> </input> <output> <port 2 type="file">output</port 2> </output> </executable> </task> <datalink> <from>C: port 2</from> <to>F: port 0</to> </link> <from>D: port 2</from> <to>F: port 1</to> </link> …. . </datalink>
Performance Evaluation (Synthetic Application on Belle Data Grid) Program Input 1 Input 2 Output type Task type A parameter file ycalc file parameter file C ycalc file parameter file D ycalc file parameter file E addcalc file F addcalc file G 103 parameter B Experimental Workflow xcalc addcalc file H merge three input files into one file Workflow Task Application
Test-bed Node belle. cs. mu. oz. au 4 CPU, 2 GB RAM, 70 GB HD, Melbourne RH Linux 8. 0 , Globus 2. 4 belle. anu. edu. au 4 CPU, 2 GB RAM, 70 GB HD, Canberra RH Linux 7. 3, Globus 2. 4 belle. physics. usyd. edu. au 4 CPU, 2 GB RAM, 70 GB HD, Sydney RH Linux 7. 3 , Globus 2. 4 gilels. cs. mu. oz. au 104 Machine Detail Location 1 CPU, 512 MB RAM, 10 GB HD, RH Linux 8. 0, Co. G 1. 1 Melbourne
Execution Progress 105
Comparison of Sequential and Distributed Execution Task Node Start time (min) End time (min) Time Task A belle. cs. mu. oz. au 0 4. 137 m A 3 m 59. 849 s B belle. cs. mu. oz. au 4. 169 8. 822 4. 652 m B 3 m 59. 997 s C belle. anu. edu. au 4. 174 9. 66 5. 486 m C 4 m 59. 997 s D belle. physics. usyd. edu. au 4. 281 10. 684 6. 403 m D 5 m 59. 997 s E belle. anu. edu. au 9. 669 10. 097 25. 62 s E 4. 996 s F belle. physics. usyd. edu. au 10. 708 11. 145 26. 16 s F 5. 996 s G belle. cs. mu. oz. au 10. 688 11. 152 27. 78 s G 5. 996 s H belle. cs. mu. oz. au 11. 172 11. 394 13. 32 s H 0. 005 s 0 11. 394 m Total 19 m 13 s WFEE Execution Time Distributed Execution time on Grid Testbed 106 Time Sequential Execution Time
Presentation Outline n Part 1: Introduction to Grid Computing and Applications n n n Part 2: Grid Economy and Service Oriented Computing n n n Compute Grid Application e. Science Application – Belle Analysis Data Grid Part 5: Closing Remarks n n n 107 Grid Market Directory, Grid. Bank, VPM, Grid Service Broker, G-Monitor Part 4: Performance Evaluation on the World-Wide Grid n n Challenges Service-Oriented Grid Architecture (SOGA) Realisation of SOGA Part 3: Global Grids and Gridbus Technologies n n Technology Evolution and Application Drivers Grid Challenges, Approaches, and Architecture Open Challenges in Grid Economy Analogy to Electric Power Grid Summary and Conclusion
Alessandro Volta in Paris in 1801 inside French National Institute shows the battery while in the presence of Napoleon I Fresco by N. Cianfanelli (1841) (Zoological Section "La Specula" of National History Museum of Florence University)
What ? !? ! Oh, mon Dieu ! This is a mad man… 109 …. and in the future, I imagine a Worldwide Power (Electrical) Grid …. . .
2005 - 1801 = 204 Years 110
(5) IT services as the fifth utility (water, electricity, gas, telephone, IT) 111 e. Science e. Business e. Government e. Health Multilingual e. Education …
Summary and Conclusion n Introduced requirements for an e. Science application Demonstrated suitability of Grid computing as Cyberinfrastructure for e. Science and e. Business. Grids exploit synergies that result from cooperation of autonomous entities: n n Grids allow users to dynamically lease Grid services at runtime based on their quality, cost, availability, and users Qo. S requirements. n n 112 Resource sharing, dynamic provisioning, and aggregation at global level. Delivering ICT services as computing utilities. Grids offer enormous opportunities for realizing e. Science and e. Business at global level.
Any Questions ? Web - http: //www. gridbus. org 113
Thanks for your attention! We Welcome Cooperation in Research and Commercialisation! http: /www. gridbus. org | http: //www. gridbus. com 114
Скачать презентацию Grid Computing and The Gridbus Toolkit Creating and
fa035738e9d5b83b4143b95e5f09c54a.ppt