Grid-based interoperability of workflow systems Moustafa Ghanem, InforSense Ltd Nabeel Azam Mike Boniface InforSense Ltd IT Innovation
Background: SIMDAT Developing and applying Grid technology to support Distributed, Cross-organizational, product and process development Four sectors of international economic importance: Automotive Pharmaceutical Aerospace Meteorology Seven Grid-technology development areas: Grid infrastructure Distributed Data Access VO Administration Workflows Ontologies Analysis Services Knowledge Services
Example Auto design workflows
Example Aero design workflows Aerodynamics (BAE) Design (UoS) Compute (UoS)
S7 S8S9 S4S5S6 S1S2S3 SOA and Workflows: A programmer’s view ToolsData Services Service Provider IService Provider IIService Provider III ToolsData ToolsData Organization IOrganization IIOrganization III
Key challenge for workflow in SIMDAT There are 26 partners in SIMDAT At least 3 workflow systems in use –InforSense KDE –Taverna/Freefluo –LMS Optimus –Other approaches include Various hard-coded wf applications BPEL Agreeing/Using a common workflow system (or language) is difficult –Migration may require huge re-investments –Continued debate on open source vs. proprietary In general, virtual organizations are dynamic –What happens when we add new partners? Capability Providers Grid Technologists End Users SIMDAT partners
Workflow Systems: Are they really similar?
Workflow systems quick comparison Similarities –All have drag-drop GUIs –All use XML-based languages –Similar conceptual model Differences –Different language paradigms/semantics Pure Data Flow vs. Control Flow –Iterations (not available, implicit, explicit ?) Data types and operators supported –Varying degrees of abstraction Node granularity Service call vs. abstract task –Different systems specialized to particular applications Specialized viewers Specialized helper nodes for data manipulation
Grid-based interoperability of workflow systems
Calling Grid Services vs Web Services Call Grid Service Call Grid Service Call Grid Service Remote Grid Service
Example: Using of GRIA services GRIA uses a fully decentralized management approach, with minimal dependency between sites. Each site offering GRIA services makes its own business decisions about which users to trust and on what terms, and is responsible for enforcing its own access policies and deciding which applications to support. Sites can interact with each other, but this is driven by their common consumers, and those consumers are responsible for managing the resulting dependencies. There are no global agreements to set up, and no virtual organizations need be established, though users can interact according to virtual organization models if they want.
Example: Pharma application Five basic Sub-workflow nodes 1.Allocate Resources 2.Upload Inputs 3.Execute 4.Retrieve Result 5.Finish Conversation
Grid-based workflows and interoperability Run-time interoperability –Workflow-based applications as services –Workflow enactment as services
Method 1: Workflow-based application as service
Example: Aero application
Method 2: Workflow enactment engine as service
Interoperability interaction patterns (a) Chained Process Model(b) Nested Sub-process Model (c) Parallel Synchronized Model(d) Polling Synchronized Model
Current Status
Next Steps: Moving from abstract to heterogeneous executable workflows
Overall approach User specifies high-level steps Steps iteratively instantiated using services and workflow templates Use service registries and workflow warehouses Use existing run-time interoperability mechanisms
Architecture Heterogeneous workflow warehouse –InforSense Workflow warehouse Service Registries –NEC Semantic Broker Authoring advisor: high-level steps for application domains Authoring assistant: map from abstract tasks to workflow fragments and services
Workflow warehousing and mining
Summary Using industrial strength workflow systems for cross-organization product design Run-time interoperability mechanisms Supporting the design of heterogeneous workflows