Holding slide prior to starting show

1
Holding slide prior to starting show
2
Grid Projects at WeSC Synergies and
Opportunities

• David W. Walker
• School of Computer Science
• Cardiff University

http//www.cs.cf.ac.uk/User/David.W.Walker/
3
Overview of Activities

• Software development of middleware, tools, and
  problem-solving environments.
• Triana UDDIe JACAW
• SWFL/JISGA G-QoSM MEDLI
• Funded research projects (RCEU)
• GSiB WOSE BD-World
• GridLab PASOA e-HPTX
• GridOneD GENSS
• Five collaborative industrial projects (DTI).
• Also work in patterns and operators, performance
  and evaluation, semantic web technologies, and
  Grid economies.

4
Collaborative Industrial Projects

• Constraint-Oriented Negotiation in an Open
  Information Services Environment (CONOISE-G).
  Alex Gray
• Collaborative Virtual Teams (COVITE). Alex Gray
  and John Miles
• Environment for Industrial Design Optimisation
  (DIPSO). Omer Rana
• Grid-Enables Computational Electromagnetics
  (GECEM). David Walker
• Resource-Aware Visualisation Environment (RAVE).
  David Walker

5
Staff Researchers in COMSC

• Nick Avis Medical imaging, collaborative
  visualisation.
• Alex Gray, Andrew Jones, Jianhua Shao
  Bio-Informatics, information/knowledge management
• Yan Huang Jini-based Grid middleware, workflow
  description, composition, deployment and
  enactment
• Omer Rana QoS frameworks, provenance and
  metadata issues, agent technologies
• Ian Taylor APIs for Grid computing, workflow,
  composition.
• David Walker PSEs/portals, workflow,
  visualisation and Grid applications.

6
Other Cardiff Researchers

• Sathyaprakash (PHYSX)
• Peter Kille (BIOSC)
• John Miles (ENGIN)
• Also interest in PSYCH,ENCAP, ARCHI, EARTH, and
  UWCM.

7
Grid Middleware

• Lightweight service-oriented architecture for
  grids based on Jini and P2P technologies.
• Workflow tools and description languages
• Grid execution environments
• Quality of service frameworks
• Provenance and other metadata issues
• Collaborative visualisation and collaborative
  working

8
Web Services

• Everything is a (Web/Grid) service.
• This includes
• Computation routines
• Access to files and databases
• Components of the Grid infrastructure, such as
  workflow enactment engines, resource monitors,
  etc.

9
A Common Approach to Workflow

• Visual service composition.
• Service interfaces and other metadata expressed
  in an XML-based service description document.
• Services registered with, and discovered through,
  a registry.

10
Workflow Within an SOA

• Workflow description
• SWFL
• Visual composition of services
• Triana
• Aggregation
• Higher level services and applications
• PSE (or portal) for deploying. Managing,and
  monitoring services, applications and grid
  resources.
• GSiB

11
Workflow in GSiB and JISGA
swfl Job
Workflow Engine
Drawing Screen
Create SWFL
Validator
Execution Environment
Visual Service Composition Environment
12
SWFL2Java

• It converts the description of a job in SWFL into
  executable Java code.

SWFL Description Document
Java Executable Code
Dataflow Graph
ControlFlow Graph
FlowModel
13
JISGA

• JISGA consists of two main parts
• A WorkflowEngine service
• A JobProcessor service
• Grid application is submitted to a WorkflowEngine
  service as SWFL.
• Sequential jobs are handled directly by the
  WorkflowEngine service.
• Parallel jobs involve multiple JobProcessor
  services.

14
What the Workflow Engine Is

• Processes service-based applications described in
  an XML-based workflow language.
• Determines the order of execution and generates
  the harness code.
• Executes the code which includes discovering
  services, invoking services, receiving results
  and sending results to clients.

15
What the Workflow Engine Does

• Converts a SWFL description of a composite
  service-based job into an executable Java code,
  and executes it.

To be continued
16
WOSE

• Workflow Optimisation Services for e-Science
  Applications.
• Middleware Open Call.
• Collaboration with Imperial College and Daresbury
  Lab.
• 400k, 2 years, one postdoc at each site
• Status
• Advertising for postdocs
• 1 Dec 2003 start date

17
WOSE Overview

• Draws together JISGA and Triana work at CU, with
  ICENI at IC, and portal expertise at DL.
• Topics addressed
• Service aggregation and deployment
• Runtime discovery and late binding of services
• Service discovery and selection from multiple
  semantically equivalent services

18
Quality of Service Framework

• Service discovery using QoS properties
• Guarantees QoS at the application, middlware and
  resource levels (similar to DiffServ), and
  establishes Service Level Agreements
• Support for QoS adaptation
• Implementation using GARA/DSRT, NRM/Diffserv BB,
  and UDDIe
• UDDIe supports the description of a service
  through service properties, and service discovery
  based on these properties.

19
Provenance and Metadata

• Important in many middleware and application
  projects.
• Two main middleware projects
• PASOA Provenance-Aware Service-Oriented
  Architecture
• GENSS Grid-Enabled Numerical and Symbolic
  Services
• Also key in BD-World application project.

20
PASOA

• Provenance-Aware Service-Oriented Architecture.
• Fundamental Computer Science for e-Science call.
• Collaboration with Southampton Univ.
• 443k, 3 years, one postdoc and student each.
• Status
• Recently funded
• Aiming for 1 Feb 2004 start date.

21
PASOA outline

• Execution and service provenance in relation to
  workflow enactment.
• Algorithms to reason over provenance data,to help
  scientists to achieve better utilisation of Grid
  resources for their specific tasks.
• Generating provenance data in workflow enactment.
  
• Properties that can be deduced from
  provenance-based data.
• Prototype that supports provenance generation and
  reasoning in Grid environments.

22
Collaborative Visualisation

• Central to two joint industrial projects
• RAVE Resource-Aware Visualisation Environment.
• GECEM Grid-Enabled Computational
  Electromagnetics.

23
RAVE Project

• Resource-Aware Visualisation Environment
• Status
• Started 1 April 2003.
• Collaboration agreement in place.
• Partners SGI and ORNL
• Duration 3 years
• Partner contribution 150,000 (SGI)
• EPSRC/DTI contribution 186,534
• Staff Dr Ian Grimstead hired as postdoc.

24
RAVE Overview

• Aims to develop a collaborative visualization
  environment that scales across a wide range of
  network-enabled devices.
• Will respond to changes in network bandwidth and
  capabilities of the target display device.
• Will start by examining VizServer and COVISE
  systems

25
GECEM Project

• Grid-Enabled Computational Electromagnetics
• Status
• Start date 1 May 2003.
• Collaboration agreement not in place yet
• Partners Swansea University, BAE Systems,
  Hewlett-Packard, and Singapore Institute of High
  Performance Computing
• Duration 2 years
• Partner net contributions 113,750 (BAE
  Systems), 113,750 Hewlett-Packard
• EPSRC/DTI contribution 227,500
• Staff Postdoc in place at Swansea waiting for
  work permit for CU postdoc

26
GECEM Overview

• Aims to use and develop Grid technology as an
  enabler of large-scale and globally-distributed
  scientific and engineering research.
• The focus of the project will be collaborative
  numerical simulation and visualisation between
  the UK and Singapore.

27
Two Hard Problems

• Semantic specification of applications
• Scheduling of workflow nodes on distributed
  resources.
• Early binding model bind to specific
  service/platform at composition time
  (validation).
• Intermediate binding model bind at compile
  time (when converting from XML to executable
  form).
• Late binding model bind dynamically at runtime.
• Later binding allows the use of more up-to-date
  information to make scheduling decisions.

28
Key Research Problems

• Semantic specification of applications
• Scheduling of workflow nodes on distributed
  resources.
• Early binding model bind to specific
  service/platform at composition time
  (validation).
• Intermediate binding model bind at compile
  time (when converting from XML to executable
  form).
• Late binding model bind dynamically at runtime.
• Later binding allows the use of more up-to-date
  information to make scheduling decisions.
• How to deal with volatile services
• Also need to discuss limitations of workflow
  approach.

29
A WeSC Service Repository

• We need to create a service repository for the
  publishing and discovering services created in
  WeSC projects.
• Initially based on UDDI 3.0.
• Will allow projects to make use of and experiment
  with services developed by other projects.

30
Summary of Activities

• Lightweight Grids.
• Visual Service Composition Environment for
  creating services and applications based on
  workflow
• Workflow enactment and execution environments.
• Collaborative visualisation.
• Quality-of-Service.
• Provenance and metadata.

31
(No Transcript)