M. Adorni, F. Arcelli, D. Ardagna, L. Baresi, C. Batini, C. Cappiello, M. Comerio, M. Comuzzi, F. De Paoli, C. Francalanci, S.Grega, P. Losi, A.Maurino, S. Modafferi, B.

Pernici, C. Raibulet, F. Tisato

The MAIS approach to web service designThe MAIS approach to web service design

Porto 13 June 2005

Andrea MaurinoAndrea Maurino

Università di Milano BicoccaUniversità di Milano Bicocca

maurino@disco.unimib.itmaurino@disco.unimib.it

2The MAIS approach to web service design

OutlineOutline

The MAIS Project

The MAIS Methodological Framework • General view

Service Specification and Compatibility Analysis Broker-Provider Negotiation and Dynamic Evaluation of

Management Costs Process Partitioning Optimal Service Selection and Quality Renegotiation Implementation Guidelines for a QoS-Oriented Reflective

Architecture

Conclusions and future work

4The MAIS approach to web service design

The MAIS ProjectThe MAIS Project

Develop methods, models, and architectures for multichannel adaptive information systems

Multichannel• Service provisioning: web, e-mail, sms, client-server

• Mobile information systems

Heterogeneity, dynamic evolution of channels• Device characteristics

• Network connections

Adaptation to context• User model

• Service provisioning

• Channel of invocation, QoS

5The MAIS approach to web service design

Research focResearch focusus

Adaptive orchestration of e-services Reflexive architecture for context-aware services Design of web Services with QoS

Adaptive networks Low power consumption processors

Methods and tools for multichannel interface design and integration

Application areas: e-learning, tourism, emergency teams

7The MAIS approach to web service design

The MAIS methodological frameworkThe MAIS methodological framework

4 Methodological steps

Requirements Analysis

• elicit, validate and negotiate web service requirements

• Multichannel, QoS

Design

• model services with MAIS-UML

Deploy

• Network infrastructure

Run time

• adaptive and context-aware use of web services

8The MAIS approach to web service design

FrameworkFramework

Analysis

Design

Deployment

Runtime

Service specification and compatibility analysis

Broker-provider negotiation and dynamic evaluation of management costs

Process Partitioning

Optimal service selection and quality renegotiation

Broker-provider negotiation and dynamic evaluation of management costs

Input/output

Interdependence

Domain Req.QoS Req.User Req.Actual ServicesArchitectural Req. UML Diagrams

QoS Req.User Req.PriceArchitectural Req.Budget

Architectural Configuration

Domain and QoS Req.Global/Local ConstraintsPriceMAIS-Registry

Process Description

9The MAIS approach to web service design

Service Specification and compatibility analysisService Specification and compatibility analysis

Analysis, (re)-design of web service

• Use of MAIS QoS, Services and Channel Ontology

3 sub-phases:

functional service modelling,

• Model requirements thanks UML

• Logical and operational structure of Web Services

high-level redesign

• Augment existing schema with QoS dimension (from the MAIS QoS ontology)

• Modelled with MAIS-UML profile

• Metrics QoS dimensions have a (Bk)) quality level

10The MAIS approach to web service design

Service Specification and compatibility analysisService Specification and compatibility analysis

Context adaptation

• Verification of quality values of QoS requirements and Quality thresholds

Comparison between Bk of each QoS and ideal level

request by user

QoS tree

Simple Additive Weighting technique

• Weights and composition rules to each node

11The MAIS approach to web service design

Process PartitioningProcess Partitioning

New deployment strategies for MobIS Goals:

• improve the independency among actors

• Reduce interaction and knowledge sharing

Solution: decompose a unique process into a set of sub process• Automatic partitioning rules

• Based on graph transformation systems

Input: MAIS-UML (translated into MAIS-BPEL), network topology

Output: set of MAIS-BPEL Some formal and empirical result on the correct

behaviour of sub process

12The MAIS approach to web service design

Optimal Service Selection and Quality RenegotiationOptimal Service Selection and Quality Renegotiation

Goal: select from registry a set of services • functional equivalent service

Optimization problem with two difference approach Select at run time the best candidate service which supports

the execution of a running high level activity. (local level)• Service can be selected if its price is lower than a given threshold

Identiy the set of candidate services which satisfy end-user preferences for an entire application (global level)• the total price has to be less than 2$.

Service composition with QoS modelled as a mixed integer linear programming problem • NP-hard !!! Multiple Choice Multiple Dimension Knapsack problem

• We use CPLEX, a state of the art commercial solver, which implements a branch and cut technique

13The MAIS approach to web service design

Broker-Provider Negotiation and Dynamic Evaluation of Broker-Provider Negotiation and Dynamic Evaluation of Management CostsManagement Costs

Broker between provider and users Conflicting Goals

• Maximize the satisfaction of user requirements

• Achieve maximum possible returns from its brokering role

Broker is paid each time a service is supplied by user Broker can increase the QoS offered by provider Preliminary phase

• set the value of a triple <pij,percij,qij>

• pij :price paid by the user for the service,

• percij :is the percentage on the price due by the service

provider to the broker (0 and 1)

• qij is the aggregate value of QoS offered to user (0 ≤ qi j≤ 1).

14The MAIS approach to web service design

Broker-Provider Negotiation and Dynamic Evaluation of Broker-Provider Negotiation and Dynamic Evaluation of Management CostsManagement Costs

Negotiation processes is realized by means of • Negotiation Protocol:

• bilateral bargaining protocol;

• Negotiation Objectives: • typical multiattribute problem: negotiation of a triple of attributes

• Decision Model: • trade-off based strategy

Utility function V • evaluate how much an offer is worth to a participant

Increase QoS level an extra cost c*(qij),

• provide the service to the customer at a higher price p*(qij)

Modify selection of web services in case of cooperative process

Even in the design phase

15The MAIS approach to web service design

Implementation Guidelines for a QoS-Oriented Reflective Implementation Guidelines for a QoS-Oriented Reflective ArchitectureArchitecture

Architectural reflection introduces a reflective layer

• applications can observe and control at execution time non-functional features of the system components,

• supporting adaptability.

A reflective layer is causally connected to the physical layer.

Components and QoS cannot be defied in absolute way

• domain-dependent QoS like “low”, “medium”, “high”

Definition of QoS extension pattern

16The MAIS approach to web service design

Implementation Guidelines for a QoS-Oriented Reflective Implementation Guidelines for a QoS-Oriented Reflective ArchitectureArchitecture

R_Object

getQoS()setQoS()

QoSQuantitativeQoSQualitative

QoS

name : StringunitOfMeasure : String

QoSValue

QoSValueSet

1

1

1

1

1

1..*

1

1..*

1

1

+actualValue

1

1R_ElementalR_Aggregate 1 0..*1 0..*

QoSStrategy

mapUp()mapDown()

1

1

1

1

R_Object expose measurable QoS values

• Controlled/observed

• Model high-level/domain-dependent abstraction

R_aggregate casually connected to QoSStrategy to set of QoS

QoSStrategy

• How QoS is obtained/mapped

17The MAIS approach to web service design

ConclusionConclusion

Propose of a methodological framework

• From Analysis to run time execution

• Fusion of several contributions

Future works

Evolution of each component according to the specific research problem

Extension of our proposal to include MAIS results for the front-end

• Integration

• Validation