Xuebiao Xu

1
BioDiversityWorld (1) From Concepts to
Implementations

• Xuebiao Xu
• Cardiff University
• 08/12/2004

2
Goal of The BDWorld

• Build a flexible, interoperable and extensible
  PSE in a Grid environment to enable biologists to
  perform analysis procedures consisting of complex
  sequences of operations involving the acquisition
  of data sets,, their analysis, and visualization.

3
Project People

• PIs
• Professor Frank Bisby, The University of Reading
• Professor W.A.Gray, Cardiff University
• Dr Andrew.C. Jones, Cardiff University
• Dr Richard J. White, Cardiff University
• Dr Alastair Culham, Reading University
• Professor Nick Fiddian, Cardiff University
• RAs
• PSE (Problem Solving Environment ) Team
• 3 RAs Cardiff UniversityMikhaila Burgess,
  Jaspreet S. Pahwa, Xuebiao Xu
• 1 RA in Southampton Oliver Bromley
• Application Team
• 4 RAs in Reading University Neil Caithness, Tim
  Sutton, Chris Yesson, Peter Brewer
• 1 RA in Natural History Museum Shonil Bhagwat

4
BDWorld Part

• Communication Layer ---Bdworld Grid Interface
  (BGI)
• To isolate the implementation from the
  invocation
• Provide language independence
• Opaque in Application Level
• Unique/uniform Invocation mechanism for all
  resources and wrappers
• InvokeOperation (ResourceHandler, Operation,
  DataCollection)

5
BDWorld Grid Interface(BGI)
BDW GRID INTERFACE (BGI)
BGIWrapper
AbstractWrapper
MDR Wrapper
Concrete W1
Ontology Wrappers
Concrete W2
.
RMI Wrapper
SOAP Wrapper
OGSA Wrapper
P2P Wrapper
WSDL Wrapper
6
3 Existing Prototypes up to now

• Standalone Java based Prototype
• Web based (Fixed) Workflow
• Triana enhanced Workflow Prototype ( Building
  Units in Static Toolbox)

7
Triana and GAT
Triana is a visual programming environment that
represents operations as boxes (units)
connected by data flow pipes, thus allowing users
to create complex computer programs out of basic
building blocks that can be dragged into a
work-space window and connected up using the
mouse, with a minimum of effort and coding.
Figure 2. Triana using the GAT
8
Triana into BDW Architecture(1)
Triana Visualization Environment
Triana Workflow Enactment System
GAP
Non-GAP
Specific Unit Builders
GAP Local Client
Triana Relevant Packages
Pipe
Peer
GAP Server
Distribution
Distribution Handler
GAP Distribution Mechanism(GAP Engine)
InitDistribution
GATDistribution Adapter
GATDistribution Event
GATDistribution Factory
Distribution Core
GAT Interface
Function Binding
BDW Init Dist. Proxy For Triana
BDW Dist. Event Proxy For Triana
BDW Dist. Factory Proxy For Triana
BDW Dist.Misc Proxy For Triana
BDW Dist. Adaptor Proxy For Triana
BDW Relevant Packages
Stub
Stub
Stub
Stub
Stub
BDW Generic Workflow Platform
Figure 3. General framework for the GAT based
Integration of BDW with Triana Workflow Enactment
System
9
Triana Into BDW Architecture(2)
BDW Misc. Proxy
Future Enhanced Workflow Manager/Hub ( i.e,
Kepler)
BDW Misc. Proxy
BDW Init Dist. Proxy
BDW Adaptor. Proxy
BDW Factory Proxy
BDW Adaptor Proxy
BDW Factory Proxy
BDW Event Proxy
BDW Init Dist. Proxy
BDW Event Proxy
Triana Related Hub
Generic Workflow Interface From BDW(GWIB)
BDW Generic Workflow Platform
Metadata Agent
Ontology Manager
Ontology Manager Client
Monitoring Component
BGI based Workflow Core
MetaData Agent Client
BGI Wrapper Proxy
BDW GRID INTERFACE (BGI)
BGIWrapper
AbstractWrapper
MDR Wrapper
Concrete Wrapper1
Ontology Wrappers
Concrete Wrapper2
.
RMI Wrapper
SOAP Wrapper
OGSA Wrapper
P2P Wrapper
WSDL Wrapper
10
Enhancements for Triana into BDW Architecture(1)

• Embedded resource locator within Triana (based on
  MA client)
• Embedded resource matcher within Triana (based on
  MA client )
• Templates for Unit Writer classes
• Unit conversion for input/output nodes
• Embedded MDR Client within Triana based
  Visualization Environment (or separated from
  Triana UI ?)
• Browsing/Editing of MDR within Triana based
  Visualization Environment.

11
Enhancements for Triana into BDW Architecture(2)

• Exception and error processing
• Knowledge based domain client for ontology
  managers to make intelligent decision.
• Multiple types supporting of input/output streams
  (part of, is-a relationship)
• Changes to GAT (Grid Application Toolkit) from
  GAP(Grid Application Prototype).
• Control Flow Hybrid with Direct Data Flow.