dynamic adaptation of service compositions using service offerings presenter: vladimir tosic (work...
TRANSCRIPT
Dynamic Adaptation of Service Compositions Using Service
Offerings
Presenter: Vladimir Tosic (work with K. Patel, B. Pagurek, B. Esfandiari)Network Management & Artificial Intelligence
Lab Department of Systems & Computer Engineering
Carleton University, Ottawa, Canada
Outline
Introduction to service-based software systems Overview of goals, definition of terminology, and an
example
Our research Research goals and our approach Service offerings and WSOL Dynamic adaptation capabilities The example revisited Current status and future work The DAMSC infrastructure
Conclusions; questions and answers
Service-Based Software Systems
“Software is a service” business modelApplications decomposed into distributed
components - services
Different terminology: service components, Web Services, e-services, dynamic services, eUtilities, …
Web Services industrial initiatives from Microsoft, IBM, HP, Sun, Oracle, …
Standards by W3C (World Wide Web Consortium)
Some Goals of the Work on Web Services
Faster development of complex software systemsFlexible, adaptable, and agile business
relationshipsAd hoc A2A (Application-to-Application) and B2B
(Business-to-Business) integration
Interoperability of heterogeneous Web ServicesLeveraging existing Internet/Web infrastructure;
based on XML (Extensible Markup Language)“Scalability to the Web”
=> Application area: e- and m-business systems
Definition of aService Component
A unit of service provisioning and management Encapsulates some service functionality and
appropriate dataSoftware-based and/or hardware-based
Composable, reusable, and replaceableMain use - service compositionsSupplier and consumer
A special case: Web Service
Example
An m-commerce system built from third-party pay-per-use service components
UserManager
ProductInfoRepository
ProductStockManager
OrderManager
Server(includes ShoppingCart states)
ClientInteraction
CreditCardVerification
Our Research Goals
Service compositions have to be managedDynamic and autonomous adaptation is one
management activityApproaches based on finding alternative service
components are not always appropriate
We work on additional and complementary support within service components to dynamically and autonomously adapt the composed
service components and their relationships, without breaking these relationships
Our Approach
Dynamic adaptation using manipulation of classes of service Introduce the concept of classes of service for
service components (extrapolate the corresponding concepts from telecommunications)
Develop appropriate adaptation capabilities & algorithms
Implementation-independent solutions and their mapping to Web Services
Appropriate infrastructure and its prototypes
Multiple Classes of Service
Variations of service and QoS (Quality of Service)
Same functionality, but differ in constraints functional constraints non-functional (QoS) constraints authorization policies cost …
Increase chances to succeed in the market by broadening the market segment
Enable balancing limited underlying resources, as well as the price/performance ratio
A Service Offering
A formal specification of a class of serviceAllowed combinations of interface-level service
offerings determine component-level service offerings
Specified separately from the specification of functionality
Constraint dimensions are separated for easier specification, but integrated into service offerings for easier choice by consumers
Specification of relationships between service offerings is a very important issue
Service Offerings vs. Potential Alternatives
Various potential alternativesService offerings are an additional &
complementary mechanism for differentiation of service and QoS
Not a complete replacement for alternatives
Advantages of service offerings include: additional flexibility relatively low overhead limited complexity of required management support for dynamic adaptation mechanisms
WSOL - Web Service Offerings Language
Explicit and formal specification of various constraints and service offerings for Web Services in XML supports dynamic selection of Web Services minimizes unexpected feature interactions
Extends WSDL (Web Services Description Language)
We are working on prototypes for: a WSOL parser with syntax and some semantic checks (automatic generation of WSOL files using Java) (Java code generation from WSOL and WSDL specifications)
Reusing and extending existing Web Services tools
WSOL - Current Status& Future Work
WSOL prototype parserCurrent issue: functional constraints that
include complex expressionsFuture work: other constraints, particularly
different dimensions of QoS
Separation and integration of constraint dimensions is the main open issue
Relationships between service offerings will be specified using a combination of constraint dimensions and roles
Motivation for Dynamic Adaptation Capabilities
Goal: accommodate changes without breaking relationships between service components
Based on dynamic manipulation of service offerings
Cannot completely replace finding alternative service components, but a useful complement
Advantages include: adaptation speed enhanced robustness of relationships between service
components simplicity and relatively low overhead
ThreeDynamic Adaptation Capabilities
1. Switching between service offerings Initiated by the service component or its consumers
2. Deactivation/reactivation of service offerings With automatic accommodation of affected
consumers
3. Creation of new appropriate service offerings Not creation of new functionality The service component decides when it is possible
and allowed
The M-Commerce System Example Revisited
Benefits of service offerings and the suggested dynamic adaptation mechanisms
UserManager
ProductInfoRepository
ProductStockManager
OrderManager
Server(includes ShoppingCart states)
ClientInteraction
CreditCardVerification
DAMSC Infrastructure
DAMSC - Dynamically Adaptable and Manageable Service Compositions
Goal: to support service offerings the suggested dynamic adaptation capabilities sessions (supplier-consumer associations that
remember state of the interaction; handle persistent interactions)
Infrastructure inside service components Complements works that suggest management
infrastructure outside service components
A proof-of-concept prototype - for Web Services
DAMSC - Current Status& Future Work (Part I of II)
Standardized management operations/interfaces Investigating use/extension of the Common
Information Model (CIM)
Support for service offerings includes: storage of service offerings inside service components informing consumer about active and available service
offerings storage of relationships with other service components
Support for dynamic adaptation includes: storage of relationships between service offerings storage of rules for creation of new service offerings
DAMSC - Current Status& Future Work (Part II of II)
Dynamic adaptation will be supported in sessions It should not destroy the session state
Support for sessions Standardization of session-management operations A session element stores information about a session
at the supplier side It should also contain the constraint-checking code Automatic generation of session elements (except
code!) A number of open issues remain
Relating DAMSC to other approaches for managing service components and service compositions
Some Related Work
Web Service industrial initiatives and standards; management approaches in these initiatives
Adaptation by finding alternative service components
Network/system/application/service management
Differentiated services and classes of service in telecommunications and TINA (Telecommunications Information Networking Architecture
Formal specification of constraints for software
Conclusions (Part I of II)
This research is about improving flexibility, adaptability, and agility of service compositions
The suggested capabilities complement prior works More appropriate than alternatives in some
circumstances
Main advantages: adaptation speed enhanced robustness of relationships between service
components simplicity and relatively low overhead limited complexity of required management
However, our approach has limitations
Conclusions (Part II of II)
The research has theoretical and practical relevance The overall problem has not been previously addressed Practical relevance - e.g., for compositions of e- and m-
business Web Services
Critical analyses are conducted at all research stages To determine benefits and limitations
Proof-of-concept prototypes for Web Services To check feasibility, uncover hidden issues, and
demonstrate contributions WSOL - there are already some results DAMSC - in the beginning stages of development
Questions and Answers,For More Information…
For more information… WWW site:
http://www.sce.carleton.ca/netmanage/ Contact e-mail: [email protected]