Standards for Web Service Choreography and Orchestration: Status and Perspectives

1
Standards for Web Service Choreography and
Orchestration Status and Perspectives

• Alistair Barros
• SAP Research Centre, Brisbane, Australia
• Marlon Dumas and Phillipa Oaks
• Queensland University of Technology, Australia

2
WS Composition Standards Overview
3
Status

• Quite a few standardisation initiatives
• Convergence is rather slow, de jure/de facto
  adoption slower
• BPEL being adopted, but focuses on a subset of
  service composition needs competing with
  programming languages?
• Modelling languages in the space do not provide
  specific high-level primitives for capturing
  realistic service interactions

4
Back to basicsIssues with existing efforts

• Current standards focus on exchanges of single
  messages or request/response anything above
  requires coding
• Paradigm possibly too low-level for modellers
  sequence is the fundamental construct
• Syntax/semantics intermingled no explicit
  meta-models
• Formalisation seen as an a posteriori effort
• Language design driven by syntax/semantics, not
  enough attention to the purpose of the language.

5
Back to basicsViewpoints in service composition

• Choreographies (global interaction models,
  collaboration processes)
• Interfaces (behavioural structural)
• Required interfaces
• Provided interfaces
• Orchestration (private processes for service
  implementation and/or mediation between
  required/provided interfaces)

6
Choreography example
7
Behavioural interface example
8
Orchestration example

9
Interfaces Provided vs. Required
Provided Interface (abstraction of internal
process/implementation)
Required/expected Interface (e.g. derived from a
choreography)
?

10
Adaptation/mediation example
11
choreography
. . .
required interface
Adaptation (design)
Adaptation (design)
Adaptation (execution)
provided interface
orchestration
Internal service/API
12
Putting some order Implementation-level
composition standards
13
Observations

• Choreography is mainly a design artefact ?
  Modelling / declarative language
• Interfaces are description artefacts ?
• Modelling / declarative description language
• Orchestration adaptation are implementation
  artefacts
• Workflow-style languages, DSLs,
  general- purpose or specialised programming
  languages

14
Example of a choreography in natural language
(quoted from xCBL)

• This choreography allows for a buyer to send an
  Order message to the supplier and then Cancel the
  Order either before or after receiving one or
  more OrderResponses to the Original Order.
• Once the CancelOrder is sent, the supplier should
  not respond with any OrderResponse, and only
  provide CancelOrderResponses. They must provide a
  response to the CancelOrder message. If the
  supplier rejects the cancellation of the order,
  they may then provide an OrderResponse message to
  the original Order.
• ? Modeller reasons in terms of allows (may),
  must, can only, should not

15
Research Agenda

• Collect set of scenarios library of patterns
• From the patterns, derive and formalise a
  meta-model for service interaction modelling
• Define concrete syntaxes for meta-model
  (graphical, XML-based) ? language for
  choreography/interface modelling
• Validate language using the scenarios
• Develop tool support and methods

16
Some possible starting points

• Service Interaction Patternshttp//www.servicein
  teraction.com
• ebBP Business Transaction Patterns
• http//www.unece.org/cefact/umm/ch9_patterns.pdf
• Property Specification Patterns
  http//patterns.projects.cis.ksu.edu
• RosettaNet PIPs, xCBL Choreographies, ....