india research lab presentation subtitle: 20pt arial regular, teal r045 | g182 | b179 recommended...
TRANSCRIPT
India Research Lab
© 2005 IBM Corporation
IBM Research
Towards End to End Composition of Web Services
Biplav Srivastava*IBM India Research Lab, New Delhi
at TIFR on Jan 21, 2005*Joint work with Neeran Karnik, Vikas Agarwal, Arun Kumar, Ashish Kundu, Koustuv Dasgupta and Sumit Mittal
India Research Lab
January 21, 2005 © 2005 IBM Corporation
IBM Research
Outline
Why do compositionBusiness-inspired case
Understand potential and scope: end-to-end
Highlight issues and non-issues
A Solution Approach Decouple into ‘what’ needs to be composed (logical composition) and ‘how’ (physical composition)
A walkthrough of prototype
Practical ChallengesInformation Modeling
Measuring impact
India Research Lab
© 2005 IBM Corporation
IBM Research
Customer Problem (Mobile Telephony Service Providers)
Voice service is commoditizedHigh volume but low margins
Mobile user services are the differentiator
Data services (e.g. stock quotes, banking)
Gateways to “brick-n-mortar” services (e.g. movie tickets)
Location-aware services
Need rapid service / application development and deployment
e.g., in response to competitor, or customer demand
State of the art: manual programming, integration too slow, long time-to-market huge implications on break-even point, revenue
Source: “The Value of the IBM WebSphere Application Server for Telecom”, Scott Broussard (IBM Austin), March 11, 2003
India Research Lab
© 2005 IBM Corporation
IBM Research
Creating a new service…
Specify end-user service capability
Select vendor service providers
Design theflow
Deploy theservice
Fig.1: Creating a new end-user service
New service
capabilities
New service
providers
Network / environment
changes
• Manual business process integration
• Use tools like WSAD-IE to create flows and business logic
• Deploy using a flow engine (such as MQWF / WAS Flow Orchestrator)
• Must take changes into account
| |
India Research Laboratory
© 2004 IBM Corporation
Customer Problem…
Service/content providers are often 3rd parties Telco is the intermediary for delivery of services to enterprises/consumers
Must improve ease-of-use of its software infrastructure Must optimize the utilization of its IT infrastructure
scalability under high-volume usage
Need to adopt standards-based frameworkUse Web services to build end-user services
component-based software engineering, software reuseUse semantic annotations allowing service functionality to be programmatically composedUse autonomic, decentralized orchestration of services
3rd PartyProviders
Telco Enterprise
User
User
User
| |
India Research Laboratory
© 2004 IBM Corporation
Helpline Automation Scenario
Service Specifications Input – Customer Problem Output – Status of Problem Resolution
Steps in the workflow Reporting of the problem using a Web-
based interface Selection of a helpline agent to resolve
the problem Agent should have desired level of
expertise Agent could be a desk-based expert, or a
field-based expert Notify the appropriate expert with
relevant information about the problem Update registry with resolution status
Goal is to create the service using existing component services, in an automated manner if feasible
Problem Ticket ID, Field Agent ID
Problem Ticket ID, Resolution Status
Customer ProblemProblem Reporting
Service
Agent AssignmentService
Help DeskService
On SiteService
Problem RegistryService
Problem Ticket ID
Problem Ticket ID,Desk Agent ID
Problem Ticket ID, Resolution Status
| |
India Research Laboratory
© 2004 IBM Corporation
Component Services
Telco services Location tracker, SMS, Email, Pager, Instant Messaging, Call
Setup, Distance calculator
Domain-specific Services Problem Reporting Service
Collects relevant information about a problem via customer interaction Problem Classification Service
Classifies based on level of expertise and the desired location of expert Optimal Agent Selector Service
Finds an expert on the field closest to a customer site Database Services
Customer DB, Product DB, Expert DB, Problem Registry DB
| |
India Research Laboratory
© 2004 IBM Corporation
Composed Service: Location-Based Agent Selection Service
Location-basedSelection Service
Effect: Agent chosen based on proximity to destinationOutput:Field Agent ID
Input:Customer Location,List of Field Agent IDs
Expert DBLocation Tracker
DistanceCalculator
Optimal AgentSelectorList of Agent IDs
List of Agent Mobile Nos
List of Agent Locations
List of Agent Distances,List of Agent IDs
Nearest Agent IDCustomer Location
| |
India Research Laboratory
© 2004 IBM Corporation
Understanding the Business Problem
Major impact in application integration/ management of IT assets possible ($$) Scope is end-to-end composition
Offline service creation, deployment User in the loop, ratification important Online monitoring for Quality of Service,
automatic re-provisioning Could drive composite service life cycle
Issue Scalability of composition solution Automate service creation – control flow and data flow Modeling consistent with software engineering principles
Non-issue Fully automatic service creation because people want to be in the loop when costly
and business-critical assets are involved Usage of any specific ontologies, semantic web, etc. Metadata needs to be
represented formally and be consistent with existing modeling methodologies.
Functionality (what)
Execution(how)
| |
India Research Laboratory
© 2004 IBM Corporation
Proposed Solution
On Demand Business Process Integration for mobile SPs
Automatic Service Composition Represent business processes as web services Enhance service descriptions using semantic annotations Express requirements/goals using semantic representation Programmatically compose web service functionality to achieve desired
goal
Autonomic Service Deployment and Orchestration Decentralize the composite service based on static program analysis Deploy on a distributed infrastructure Monitor its execution at runtime Optimize the deployment periodically
| |
India Research Laboratory
© 2004 IBM Corporation
System Architecture
Abstract Workflow
(Plan)
Domain Ontolog
y
Service Registry
DeployableWorkflow
ServiceSpecificatio
n
Logical Composer
Physical Composer
Execution Environment
Service CreationEnvironment
Key Components Service Capabilities Database
Information about services available in-house as well as with 3rd party providers
Telecom Ontology Domain-specific terminology
Logical Composer including Planner automated aggregation of services via
generative planning-based reasoning techniques
Physical composer Instance selection based on end to
end QoS specification
Input Requirements document for the new
service that needs to be composed
Output Deployable workflow representing a
composite service
| |
India Research Laboratory
© 2004 IBM Corporation
Logical Composition
ValidatorOntology
SNOBASE
MatchmakerFilterPlanner4J
ServiceSpecs
Abstract Workflow
ServiceCapabilities
Registry
Logical Composer Available
ServiceTypes
DomainInformation
CandidateComponent
Services
DomainInformation
CandidateComponent
Services
ValidatedSpecs
| |
India Research Laboratory
© 2004 IBM Corporation
Physical Composition
Instance Selector
BPEL Generator
AbstractWorkflow
DeployableWorkflow
Data Dictionary
WSME
Matchmaker
ServiceInstancesRegistry
PhysicalComposer
Available ServiceInstancesSelected
Instances
WSDLs for the selected Instances
Matchmaking language/rules
MatchingInstances
| |
India Research Laboratory
© 2004 IBM Corporation
Demo Walkthrough
| |
India Research Laboratory
© 2004 IBM Corporation
| |
India Research Laboratory
© 2004 IBM Corporation
| |
India Research Laboratory
© 2004 IBM Corporation
| |
India Research Laboratory
© 2004 IBM Corporation
| |
India Research Laboratory
© 2004 IBM Corporation
| |
India Research Laboratory
© 2004 IBM Corporation
Composed Service: Helpline Automation
Problem Reporting
RegistryUpdate
CallSetup
Help Desk
MessageDelivery
On Site
Problem Classification
Location-basedAgent Selection
Expert LookupAgent Assignment
Desk-based Expert ID Field Expert ID
Problem Ticket
Problem Ticket, Problem Ticket,
Resolution StatusProblem Ticket,
Resolution StatusProblem Ticket,
Customer Interaction
| |
India Research Laboratory
© 2004 IBM Corporation
| |
India Research Laboratory
© 2004 IBM Corporation
| |
India Research Laboratory
© 2004 IBM Corporation
| |
India Research Laboratory
© 2004 IBM Corporation
Practical challenges
End to end information modeling Planning gives control flow but not data flow readily. Data flow generation
needs contextual information about inputs and outputs. Possible to generate data flow but it will bloat the service specifications Will cause frequent specification changes to information model
Handling non-functional requirements, QoS constraints Scalability – OWL-S supports at web service instances level
Measuring impact How to quantify gain with a composition tool v/s manual methodology How on-demand can the system be?
| |
India Research Laboratory
© 2004 IBM Corporation
Concluding comments
Web services composition important to solve real business problems Presented application integration for a telecom provider
Need end-to-end, continual composition perspective Existing distributed computing solutions need to leveraged – CIM,
Monitoring, Execution Infrastructure, Grid, Autonomic Computing, etc.
Presented an approach and prototype tool Decouples composition into what and how phases Uses AI and distributed computing techniques collectively
Presented practical challenges
IBM Research
© Copyright IBM Corporation 2005
Thank you…