model-driven business process integration and management: a … · 2015-07-28 · and...

Model-driven businessprocess integrationand management:A case study with theBank SinoPac regionalservice platform

J. ZhuZ. Tian

T. LiW. Sun

S. YeW. Ding

C. C. WangG. Wu

L. WengS. Huang

B. LiuD. Chou

Business process integration and management (BPIM) is acritical element in enterprise business transformation. Smalland medium-sized businesses have their own requirements forBPIM solutions: The engagement methodology should be fastand efficient; a reusable and robust framework is required toreduce cost; and the whole platform should be lightweight sothat one can easily revise, develop, and execute solutions.We believe that model-driven technologies are the key tosolving all of the challenges mentioned above. Model Blue, aset of model-driven business integration and managementmethods, frameworks, supporting tools, and a runtimeenvironment, was developed by the IBM China ResearchLaboratory (CRL) in Beijing to study the efficacy of model-driven BPIM. To verify the technology and methodology,Model Blue was deployed with Bank SinoPac, a mid-sizedbank headquartered in Taiwan. A lightweight BPIM solutionplatform was delivered for Bank SinoPac to design, develop,and deploy its business logic and processes. During the eight-month life span of the project, IBM teams developed fourmajor solutions for Bank SinoPac, which also developedone solution independently. In spite of the remote workingenvironment and the outbreak of the Severe Acute RespiratorySyndrome illness, the project was completed successfully onschedule and within budget, with up to 30% efficiencyimprovement compared with similar projects. Bank SinoPacwas satisfied with the technology and methodology, andawarded IBM other projects. In this paper, we illustrate howeach key business process integration and solution developmentphase was carried out and guided by business processmodeling, together with major experiences gained. Thefollowing technical aspects are discussed in detail: a two-dimensional business process modeling view to integrate flowmodeling and data modeling; a lightweight processing logicautomation environment with tooling support; and the end-to-end BPIM methodology, with models and documentssuccessfully integrated as part of (or replacement for) thedeliverables defined in the existing servicing methodologiesand software engineering approaches.

�Copyright 2004 by International Business Machines Corporation. Copying in printed form for private use is permitted without payment of royalty provided that (1) eachreproduction is done without alteration and (2) the Journal reference and IBM copyright notice are included on the first page. The title and abstract, but no other portions, of thispaper may be copied or distributed royalty free without further permission by computer-based and other information-service systems. Permission to republish any other portion of

this paper must be obtained from the Editor.

0018-8646/04/$5.00 © 2004 IBM

IBM J. RES. & DEV. VOL. 48 NO. 5/6 SEPTEMBER/NOVEMBER 2004 J. ZHU ET AL.

649

1. Model-driven business integrationBusiness Process Integration and Management (BPIM) isimportant for transforming today�s business enterprises.Enterprises require end-to-end solutions in order toeffectively link internal and external business applications,systems, and staff so that they can respond with flexibilityand speed to changing business conditions. Thetechnologies for business integration have evolved overthe last twenty years.

Most early integration solutions were focused onconnecting systems. Vendors have provided variousEnterprise Application Integration (EAI) adapters(e.g., iWay [1]) to help create linkages among differentapplications (e.g., SAP**, SQL Server, and DB2*). Suchsolutions tend to establish relationships among systemsin an ad hoc point-to-point manner. This is illustrated inFigure 1 with applications in a typical bank as an example.Usually it leads to a complex interaction network amongsystems and business units and results in problems inareas such as performance, maintenance, lack of scalabilityand extensibility, and stability.

The concept of an integration hub (also shown inFigure 1) has been widely adopted by today�s leadingbusiness application integration solutions, such as IBMWebSphere* Business Integration (WBI) [2], MicrosoftBizTalk Server** [3], and BEA Weblogic** Integrator[4]. These solutions connect different systems througha centralized integration engine or hub, where theintegration logic resides and executes. As a result, thecreation, maintenance, and changing of integrationlogic can be managed in a more flexible and efficientway.

The most challenging part that remains unsolved,however, is how to plan, build, maintain, and utilize theintegration hub for real business cases. Many solutions arestill focused on the information technology (IT) aspects of

system integration, which, it is asserted, is becoming thecommodity portion of enterprise integration solutions [5].Historically, there have been attempts and practices toimprove enterprise businesses through IT advances, butaccording to [6], these have not proven to be verysuccessful. The differentiation of today�s businessintegration solutions is elevated to the design and analysisof high-level business strategy and processes, and the wayin which IT systems can be refined or created to supportthem. From this perspective, an integration hub isinsufficient. In fact, an integration platform is required toguide the end-to-end solution lifecycle. Such a solutionlifecycle typically includes phases such as requirementcollection, macro and micro design, implementation, test,and deployment.

Model-driven business integration (MDBI) is anemerging approach for building such an integrationplatform. The distinctive feature of this approach isthat models are employed as the key elements to drivethe end-to-end integration solutions from businessrequirements down to IT implementation. For each majorphase of a solution lifecycle, guidelines are given tospecify the input models, output models, and modeloperations (e.g., creation, transformation, andmodification). According to the layered modeling conceptin Model-Driven Architecture (MDA) [7], the separationof business logic from IT technology can be achieved byusing separate models at the computation-independentmodel (CIM) layer, the platform-independent model(PIM) layer, and the platform-specific model (PSM) layer.Model transformations among layers will help to fill theso-called business–IT gap.

As reasonable as it sounds, most work in MDBI is stillin the research stage and has not been verified by studiesthrough real business cases. The fundamental challengefor this approach is not each single modeling method and

Figure 1

Traditional Enterprise Application Integration (EAI) compared with an integration hub.

Core

banking

Risk

management

Managerial

accounting

Credit

card

International

settlement

Fund

management

Risk

management

Core

banking

Credit

card

Managerial

accounting

Fund

management

International

settlement

Integration

hub

J. ZHU ET AL. IBM J. RES. & DEV. VOL. 48 NO. 5/6 SEPTEMBER/NOVEMBER 2004

650

tool itself, but how to build a methodology with tools tofacilitate its execution. The success of this methodologycan be attributed to the careful crafting of the tools to thedifferent roles of the methodology, and the integrationof the tools via the transformation of artifacts. This isactually something that cannot be fully verified withoutfield practices. Therefore, it has been a goal of thisresearch to couple and apply the model-driven approachto real business practices and gain insight on howpractitioners, rather than just researchers and technologydevelopers, will use these approaches.

A set of model-driven business integration andmanagement methods, frameworks, supporting tools, andruntime environment were created in 2002 by the IBMChina Research Laboratory (CRL) to conduct thisstudy. For convenience, this set of technologies andmethodologies is called Model Blue in this paper. ModelBlue forms a platform with an end-to-end methodology tohelp enterprises meet high-level business goals througheffectively connecting internal and external businessapplications and systems. As illustrated in Figure 2,Model Blue includes technologies for modeling businessprocesses in two views—the Model Blue Business Viewand the Model Blue IT View. The Model Blue BusinessView provides an easy and intuitive method of modelingintegrated business processes. The Model Blue IT Viewoffers a lightweight processing logic modeling method withtooling and runtime support. The IT View model can alsobe transformed to other formats (i.e., WBI Collaboration,Flow Definition Language (FDL) [8], and BusinessProcess Execution Language (BPEL) [9]) and integratedwith related commercial tools and runtimes to form theModel Blue solution suite. The Business View and ITView are seamlessly connected with model transformersunder the guidance of the Model Blue solutionmethodology. In the remainder of this paper, the termModel Blue is used to represent the whole integratedsolution suite and methodology; the terms Business Viewand IT View, if not specifically noted, are used torepresent the Model Blue Business View and the ModelBlue IT View, respectively.

Bank SinoPac, an Asia Pacific regional bankheadquartered in Taiwan and known for its business andtechnical innovation, has the need to support its businessstrategy with a business integration platform, on which itcan design, develop, and deploy business processes for itscross-Pacific (including mainland China, Hong Kong,Taiwan, and the U.S.) financial products. Bank SinoPachas its specific requirements for BPIM: The projectfulfillment methodology should be fast and efficient; areusable and robust framework is required to reduce cost;the whole platform should be intuitive and easy to useso that Bank SinoPac can revise, design, and executesolutions independently. These requirements are still a

challenge for today�s business integration solutions. TheIBM China Research Laboratory, in collaboration withIBM Business Consulting Services (BCS), took thisopportunity to apply the Model Blue technologies tobuild the integration platform for Bank SinoPac. Onemission of the engagement was to verify the anticipatedbenefits of a model-driven approach in solving real BPIMproblems.

One distinctive advantage of a model-driven approach isthe representation of business and IT issues with formalmodels, which make it possible to apply formal analyticalmethods to derive valuable insight, for example, regardingprocess correctness and effectiveness. Basic business processmodel analytical capabilities have been developed andshipped with Model Blue (i.e., to facilitate the detectionof interprocess conflicts).

The rest of the paper is organized as follows: Part 2gives an overview of Model Blue, including its solutionsuite, Business View, and IT View. In Part 3, the casestudy is presented in detail, including the profile andrequirements of Bank SinoPac, challenges for BPIM, andthe execution lifecycle, highlighting major experiencesgained. Best practices and lessons learned are presented inPart 4. Part 5 concludes the paper with user feedback andtechnical highlights. Part 6 briefly indicates some futureresearch challenges.

2. Overview of Model BlueIn this paper, we use the term Model Blue to represent aset of model-driven business integration and management

Model-driven integration platform.

Figure 2

Legacy

application

Legacy

system1

Partner

system

Customer

system

New

App2 New

App1Legacy

system2

CustomerBusiness

analysts/

consultants

Architect

ProgrammerTester

Project manager

Model-driven

integration

platform

Req

uire

ment Design

Test Develo

p

Bus

iness View

IT

Vie

w


651

methods, frameworks, supporting tools, and runtimeenvironment. Model Blue Business View provides themodeling methods and tools to help consultants andanalysts collect, understand, formulate, discuss, andanalyze enterprise business processes. The Model Blue ITView offers a lightweight processing logic automationenvironment with tooling to help analysts, architects, anddesigners communicate and implement the businessrequirements in IT systems. With the help of a modeltransformation engine, the Business View can beautomatically transformed into initial IT View modelsto help bridge the business–IT gap. The IT View modelcan be executed on the Model Blue IT View Engine ortransformed to industry product format and integratedwith related commercial tools and runtimes to formthe Model Blue solution suite. All of the above partsconstitute the technical base for Model Blue solutionmethodology, which provides guidelines to seamlesslyintegrate them across an end-to-end solution lifecycle.From the model transformation point of view, Model Bluestarts with an integrated Business View combining flowand data requirements from business consultants, andthen generates two sorts of models for IT architects— onefocused on flow activities (Model Blue IT View), and onefocused on data elements (i.e., XML schema). Differenttools are then used to further develop these models acrossthe remaining solution phases such as development,testing, and deployment.

Model Blue suiteFigure 3 illustrates the whole suite of a Model Bluesolution in three parts: design tools, managing tools, andruntime environment.

Design toolsThe two major tools within the solution, Business ViewBuilder (targeted for business consultants) and IT ViewDesigner (targeted for IT staff) are both built on top ofthe Eclipse [10] platform and are linked together by themodel transformer. The two tools can also work togetherwith other modeling and developing tools, such as theWBI process designer and the WBI/ABE (AdaptiveBusiness Entity, a state-driven broker of multiple processflows) builder.

Managing toolsThe business process models created by the design tools,including the Business View model, IT View model, andWBI Collaboration model, are managed by the ModelBlue Asset Manager [11]. An application managementconsole is also provided to deploy, manage, and monitorthe executable flow models.

Runtime environmentThe Model Blue IT View Engine is mainly responsible forthe execution of process flows by routing through variousbusiness applications and/or transactions. Existing

Figure 3

WAS/MQ

New Apps/LogicWBI Server

Model Blue suite.

Legacy system Core banking Legacy system

Business View Builder

ADoc

Administration

CWT

B2B gateway

Analyst

IT staff

Model

Trig

ger

WBI process

designer

Model Blue IT View Engine

Model

transformer

IT View Designer

App M

gm

t

Conso

le

Monitor

Deploy

WBI/ABE

builder

Refine/design

BP

M M

gr

EX

E sc

ripts

Active process

stack

IT View

script

Business View

model

Runtime environment

Design tools Managing tools

Refine/design


652

application connectors in WBI are leveraged here toconnect to back-end legacy systems. They are linkedtogether with the IBM WebSphere Application Server(WAS) and WebSphere MQ Series [8] middleware to formthe runtime environment. The IT View Engine can alsoaccess legacy systems directly with specifically designedactions to support new applications and logic.

Business ViewModel Blue Business View is targeted for businessanalysts and consultants and is designed specifically tomeet their needs. A global view of the business processesis important for analysts and consultants to understand therequirements, yet it is often very hard to gather all relatedparties together to create such a global view. One uniquecapability of Business View is to provide localizedmodeling capability so that business analysts orconsultants can use it to interview individuals and outlinetheir understanding of the business process, by askingquestions such as “What are the steps for doing thisjob?”, “How do you make decisions to route through thesteps?”, and “How do you collaborate with others in thesesteps?”. Each interviewee contributes partial businessprocess requirements from an individual perspective.Process and data elements can be modeled in isolationand then assembled in a hybrid picture. Business Viewthen provides methods and tools to combine all of thesesubparts into a global model, which can then be used foranalysis or as guidance for future implementation. Sinceanalysts and consultants also wish to be able to showdifferent levels of details under different usage contexts,Business View must have flexible support in modelnavigation, expanding, and collapsing.

Basically speaking, formalized representation ofbusiness processes must include key modeling elementssuch as activities, execution logic, data definition, controlflow, information flow, and external entities. Model BlueBusiness View provides a two-dimensional usage model toabstract and organize these elements in one picture. Inthe vertical dimension, the internal execution logic withina business entity (e.g., business organization, serviceprovider, and person or role) is described with commonflow diagram constructs. In the horizontal dimension,interaction among business entities is represented byinformation exchange across entity boundaries. As shownin Figure 4, on the left side, execution logic for a businessentity (Clearing bank) is described inside an activity lane(container); on the right side, external entities (Senderbank and Receiver bank) are represented by collaborativeentities (person-like icons); across the border of thecontainer, cross-entity interactions are captured asinformation artifacts (round-cornered rectangle icons).

Several Business View models can be put together toform an assembled model on which the interaction among

different business entities can be defined and visualized.Figure 5(a) gives an example of two activity lanes beingconnected with a data lane between to describeinformation merging and mapping relationships. Twoinformation artifacts from different activity lanes may havedifferent data structure definitions. It is not necessary fortwo information artifacts to have exactly the same datadefinition in order to match. Meta-level structure mappingcan be specified on the information flow as a data format“transformer” between them in order to establish mapping.

The uniqueness of the Model Blue Business View modeland tooling lies in the specific design to help businessconsultants capture customer requirements. Theintroduction of information artifacts and collaborativeentities in Model Blue Business View enables businessconsultants to separately capture business processrequirements with individual departments or persons withno requirement for global knowledge. All of these modelscan then be assembled together with the data lane to forma global process picture for verification and analysis.

Figure 5(b) also demonstrates the mapping relationshipfrom the usage diagram model to an equivalent Petri Net[12]. The key value of the Petri Net theoretical foundationis that a formal method can be leveraged to analyze theBusiness View model for valuable insight. The traditionalworkflow net [13] analysis method is extended here to testthe compatibility among business processes and pinpoint

Figure 4

Model Blue Business View.

Receive transfer request

Transfer

successful?

Receive

update

transfer

information

Send

transfer

account

Receive

failed notice

Forward

failed notice

Forward transfer request

Clearing bank

Transfer

request

Forward transfer

information

Transferred

information

Failed

notice

Transferred

report

Failed

notice

Sender

bank

Receiver

bank


653

potential interprocess conflicts such as deadlock andinfinite waiting [14]. Figure 6 illustrates the relationship ofeach business process validation problem to its Petri Netrepresentation. Currently, the effect of using the Petri Nettheoretical model analysis is not satisfying enough: Onlylimited kinds of conflict can be identified, and the result ofanalysis appears to be trivial for simple processes. Furtherwork is expected in this field to study the applicability offormal model analysis to more valuable business problems.

IT ViewServing as the link between business process and ITimplementation, Model Blue IT View provides a set oflayered target-oriented modeling methods to lead businessrequirements toward IT implementation architecture

through predefined guidelines and processes in which ITimplementation details (e.g., auditing, performance, andlogging) can be introduced gradually. In contrast toprevalent workflow-like representations [15], the IT Viewmodel describes business process execution using anannotated tree structure, which comes from the nature ofsolving complex problems with a “divide-and-conquer”approach. As a result, it provides a common base forbusiness analysts, IT architects, and designers tocommunicate requirements and design systems. Theannotated tree contains three kinds of nodes: The actionnode is a leaf node to represent invocation of anexecutable software artifact (such as a Java** program,Enterprise JavaBeans** application, Web service, orRemote Method Invocation application); the sublogic

Figure 5

(a) Assembled Model Blue Business View; (b) Petri Net equivalent. (PO: purchase order; ACK: acknowledgment; RA: remit agreement.)

Send PO

Receive

PO ACK

Send

cloth

material

Receive

invoice

Inspection

Qualified

Payment

Receive

goods

Reject

goods

Receive

PO

Cancel

PO

Send PO

ACK

Receive

cloth

material

Receive

inspection

Passed

Send

invoice

Deliver

goods

Reject

goods

rejected

Accept

Receive

RA

POPO

PO

cancel

Delivery

order

Inspection

result

RA

Invoice

Delivery

order

Goods

rejected

PO ACK

Delivery

order

Inspection

sheet

Invoice

Letter of

advice

Rejected

order

PO ACK

Yes

YesNo

No

Yes No

Send PO

Receive

PO ACK

Send

cloth

material

Receive

invoice

Inspection

Payment

Receive

goods

Reject

goods

Receive

RA

Receive

goods

rejected

Receive

inspection

Receive

cloth

material

Cancel

PO

Receive PO

Send PO

ACK

Send

invoice

Deliver

goods

RA

(a) (b)


654

node is a leaf node to represent invocation of anotherannotated tree; the control node is a non-leaf node torepresent the execution logic (such as sequence, parallel,and choice) of its subnodes. The key value of theannotations tree is that it enables dynamic settingcapabilities to describe a complex processing logic withhigh flexibility and expressiveness, as verified by theworkflow pattern defined by [16].

A transformation engine is provided to transform theBusiness View model into an initial skeleton of the ITView model (as illustrated in Figure 7), to which detailedimplementation requirements can be added later by ITarchitects and designers. Starting from the root, the treeprogressively shows how the process is decomposed intomore manageable or understandable pieces, which can besublogic nodes that represent factoring of the model oraction nodes that represent the implementation of variousfacets of the model. Different levels and aspects of thetree presentation enable the business analyst andapplication developer to access and update thecorresponding information they care about. Theinformation about program interfaces, input/outputparameters, skills required, and major processing logic canbe exported as a development task guideline documentand then delivered to developers or vendors as a cleardescription of the development tasks.

The business process flow composition logic captured inthe tree structure can easily be adapted. The leaf nodescan easily be re-bound to another tree or executablesoftware artifact, as illustrated in Figure 8. An ActionRegistry is implemented to maintain and reuse the

interface definition of executable software artifacts. Withthe help of integrated tooling, one can easily performinterface definition import/export with Java source code.The structured logic representation also makes it easy toextract part of the function and implement it with anothermechanism. For example, we can extract a subtree andtransform it into another flow representation (i.e., WBICollaboration); the change is localized. The IT View alsosupports the debugging and simulation of the businessprocess flow application. This capability can verify thefunction of an application more easily, shortening thedesign-to-deployment cycle.

3. Case study with Bank SinoPac

Bank SinoPac profileBank SinoPac is a Taiwan-based commercial bank that wasfounded in 1992 with the corporate mission of becoming atrusted financial institution of the Pacific Rim. In the pastten years, it has been able to achieve the best asset qualityamong commercial banks in Taiwan and a financialnetwork spanning the Pacific. Despite its moderatesize, Bank SinoPac has received awards such as “BestCorporate Governance Company in Taiwan” fromEuromoney [17], “2003 Bank of the Year in Taiwan”from The Banker [18], and “Best Corporate/InstitutionalInternet Bank in Taiwan” from Global Finance [19]. In2002, it was merged with National Security Corporation toform SinoPac Holding, which now owns a client base ofmore than 1,000,000.

Figure 6

Mapping from business problem validation to Petri Net properties.

Starting from start state i, the control flow will

eventually reach end state o.

• Every accessible state M starting from i will

eventually lead to the end state o.

Business model

BP properly terminated

• Every control flow accessible

“Possibility” implies an executable possibility.

(No BP termination with any control

flow still accessible)

It is possible to execute every activity.

• But not infinitely executed

(Requirement for meaningful BP cannot

be verified by theoretical analysis.)

Once the flow is terminated, all resources have

been consumed.

• “Control flow accessible” means that resources arrive.

• No resources are trapped in mid-graph.

There are no dead transitions in the initial state i.

Flow fairness assumption: “A transition that is enabled

infinitely often will eventually fire.”

• All choices will eventually be made (implicitly or

explicitly), and they should not introduce an infinite loop.

Business process (BP) terminable

• Any available mid-state can be terminated.

Theoretical model


655

Bank SinoPac�s current operation includes 19business units at the company headquarters, 44 branchoffices throughout Taiwan, an offshore banking unit, onebranch in Los Angeles and one in Hong Kong, one liaisonoffice in Vietnam, and an affiliated bank in the UnitedStates, Far East National Bank, with a fifteen-branchnetwork and a Beijing representative office. To providean integrated financial service solution, Bank SinoPac hasinvested heavily in financial peripheral business, includinga finance company, a leasing company, and financialconsultation.

Bank SinoPac requirementsTo succeed in the highly competitive financialmarketplace, Bank SinoPac has made two critical strategicdecisions: One is to position itself as a cross-Pacific bankserving the greater China community; the other is to

increase its capabilities through alliance with otherfinancial institutions. Both strategies require a powerfulapplication middleware system that can connect to variouslegacy systems with flexible business logic processingcapabilities.

The legacy application middleware of Bank SinoPac wasbuilt primarily on the Enhanced Messaging and Queuing(EMQ) [20] technology, which had only limited businesslogic processing capabilities. Its scalability, reliability, andextensibility were also issues. Bank SinoPac had reviewedalternative solutions but excluded them because none ofthese solutions could provide full cycle management fromthe Business Level to the IT Level. Bank SinoPac intendedto rebuild its application middleware to acquire theadvanced capability of business logic processing. BankSinoPac wished to design, develop, and deploy its ownbusiness logic on the middleware platform. The

Figure 7

Model Blue IT View skeleton generated from Business View. (MMA: money management account; NTFXMF: new Taiwan and foreign

exchange mainframe; FundMF: fund mainframe; CSR: client service representative.)

Middleware (mutual fund purchasing)

Receive request

Account transfer

Result?

Fund purchasing

Result?

Reverse

Start

pending

process

Date

supersession

Notify result

Result?

Transfer request

Total request

Transfer response

Supersession

request

Purchase request

Purchase response

Reverse request

Reverse result

Pending

process report

Result

CSR Log

MMA

NTFXMF

FundMF

NTFXMF

NTFXMF

MMA

Mutual fund purchasing

Sequence

Receive

request

Account

transfer

Result?

(action)

Result? Notify

result

Fail Date errorTimeout Success

Fund

purchase

Result?

(action)

Result? Notify

result

FailSuccess

Result?

(action)

Timeout

Notify

result

Result?

Timeout or failSuccess

Reverse


656

application middleware would connect to various back-endand front-end systems through multiple channels, forinter/intra-business integration.

As a specific trial of model-driven business integrationtechnologies, the scope of the project included thefollowing:

● A regional service platform (RSP) for Bank SinoPac todesign, develop, and deploy its financial products basedon business process modeling, design, and execution.RSP will provide full lifecycle management from theBusiness Level to the IT Level.

● Key integration solutions for a strategic financialproduct of Bank SinoPac called Cross Pacific Account(CPA), through which the bank can providebusinessmen from Taiwan, mainland China, Hong Kong,and the U.S. with immediate and convenient fundtransfer services. CPA includes the following eight setsof business processes, grouped by five patterns forintegrating various front-ends, back-end systems, andbusiness units:

● Two-phase transaction (process that contains twoconsecutive transactions with logical dependency):Mutual Fund Purchasing and Combined Inquiry andUpdate.

● Parallel transaction (process that contains severaltransactions issued simultaneously and then merged):Account Portfolio Inquiry and Credit PortfolioManagement.

● Cross-regional transaction (process that involvestransactions to distributed mainframes): Cross-RegionalMoney Transfer and Account Settlement.

● Front-end-triggered transaction (process triggered byfront-end kiosks): MMA (Money Management Account)Combo Card. 1

● Back-end-triggered transaction (process triggered byback-end mainframes): Business Date Super-Session(daily refreshing mainframe date settings).

1 Money Management Account is a capital management service of Bank SinoPac tohelp company and individual clients to effectively manage investment throughintegrated multiple bank accounts.

Figure 8

Model Blue IT View with IT details.

PAP (Parallel Account Portfolio)

Sequence

Start transaction Parallel inquiry Finishing

Performance

monitor start

Generate

inquiry list

Taipei mainframe inquiry

Taipei collaboration

Generate response

Taipei collaboration

Generate response

Fund mainframe inquiry

Hong Kong mainframe inquiry

Hong Kong collaboration

Generate response

Generate total response

Performance monitor finish

Consolidate

Shanghai collaboration

Generate response

Shanghai mainframe inquiry


657

Challenges and success criteriaThe following points were identified by Bank SinoPac andIBM as the key project challenges as well as criteria bywhich to verify the Model Blue solution, based not onlyon formerly anticipated effects of the model-drivenapproach, but also on more specific requirements fromthe real situation:

● Lightweight integration solution For a mid-sized companylike Bank SinoPac, a lightweight integration solutionis not just a feature that is nice to have. The term“lightweight” here has multiple meanings. First, thewhole set of service methodology should be practicablefor a small project team with limited external technicalsupport. Second, the methods and tools should be easyto use so that they can be learned and applied by peoplewith a normal business/IT background under limitedtraining. Third, to keep a smaller footprint, most mid-sized companies wish to have a lightweight integrationruntime solely on top of a standard application server;on the other hand, the flexibility and performance of theintegration runtime should also be considered. In thisspecific case, for instance, Bank SinoPac wished to reuseexisting commercially available application connectors.

● Cost-effective testing and debugging The testing of anintegration solution usually involves linkage to multiplelegacy systems, of which most are critical online parts ofthe daily operation system. It is far more complex thanordinary software testing, which normally includes alimited number of components and dependencies andcan be executed in a staging environment. Testing hasbecome a major challenge for traditional black boxdebugging technologies [21]. Therefore, an integrateddebugging capability to easily step from process logic toconnector code (and vice versa) is highly preferable. Totake all dependencies into consideration, testers usuallyface the challenge that even thousands of test cases failto cover all possible scenarios of a simple integrationsolution; thus, the capability to automatically generatetest cases with acceptable function coverage becomesanother critical issue for the efficiency of testing.

● Easy-to-use communication tools Many business processmodeling tools today are based on either IT-Levelmodels or tools (e.g., UML [22] used in IBM RationalRose* [23]) or diagram-drawing environments (e.g.,Microsoft Visio**, iGrafx FlowCharter** [24], andSmartDraw** [25]). For business consultants andanalysts, additional training is required. Moreover, mostof the tools are not designed for the working style ofbusiness consultants and analysts (for example, how toformulate a global system picture through the results ofseparate interviews with individuals). Therefore, moreeffective communication tools are needed to help

establish common understanding on requirementsamong Bank SinoPac staff (from both business and ITunits), and within the IBM project team.

● Resolution of Business–IT gap One challenge for today�ssoftware lifecycle engineering approaches is how to bridgethe gap between business consultants and IT specialists,due primarily to differences in background, skills, andvocabulary. Some work has been done in the field ofenterprise architecture (e.g., Zachman Framework [26]and Popkin Enterprise Architecture [27]), but this workis not focused on solution development. Designing anIT implementation that really conforms to businessrequirements is essential for the success of the project.

● Asset reusability Reusability is a well-studied topic in theliterature [28]. Design documents and implementationcodes are intellectual assets which, if made reusable, willbring considerable value to current and future solutions.One criterion for software reuse efforts is the level ofreuse [29]. It is easier to reuse high-level models thanto reuse software code. Therefore, the integrated setof models across the phases of the solution lifecyclepromotes more effective asset reuse; this is an importantcriterion for the project.

● Distributed working team management Given the fact thatthere are different project teams residing in Beijing,Taipei, Hong Kong, and Nanjing, helping geographicallydistributed teams to understand the complete workingenvironment for their job assignment is a greatchallenge, and many lessons have been learned over theyears [30]. A widely distributed collection of individualsworking separately can easily lose track of the others�work and progress [31]. As a result, another challengeis allocation of work items to outside contractprogrammers and testers to clarify job specification,job context, and quality assurance criteria, whileavoiding disclosure of significant details of BankSinoPac�s business processes.

● Effective business/IT change management For anintegration project, there are complex dependenciesamong system components. One single requirementchange usually propagates across many othercomponents of the systems, especially in the latterphases of the project. To guarantee on-time and on-budget delivery, changes should be managed at theenterprise level and analyzed by the architect fortheir potential impact before being accepted andimplemented. Management of changes means managingthe process of changes as well as managing all artifactsof an evolving software system [32]. Both Bank SinoPacand the IBM project manager wish to have thecapability to demonstrate the possible effects of thechange requests, from the perspective of both functionand implementation impact, so that they can make thecorrect decisions.


658

● Alignment with existing service/software methodologiesExpecting workers to pick up a totally different way ofdoing a job is unrealistic, especially for service teamsthat have already endorsed a well-defined softwaremethodology, such as the Rational Unified Processes(RUP) [33]. Providing seamless mapping and alignmentto existing paradigms thus becomes a critical issue foracceptance of a new methodology, especially when theproject timeline is too tight to allow for muchcontingency for training time.

● End-to-end solution Each challenge above, in isolation, isnot a major issue, and we may already have answers forthem. There are still unresolved problems in addressingall of these concerns in a consistent and coherent wayso that the project can be guided smoothly from thebeginning to the end. Seamless connection of variousphases of a solution lifecycle with some level ofautomation is a critical issue for most service andsoftware methodologies. Most of today�s integrationprojects require significant investment in money andtime. Projects suffering delay caused by unexpectedevents are not uncommon. An end-to-end methodologyto mitigate these risks is an imperative for success.

Engagement lifecycle and outcomeFigure 9 shows an engagement lifecycle driven by models,more specifically Model Blue Business View and IT View

models. It is not mandatory that all solutions strictlyfollow the listed phases (i.e., requirement, macro design,micro design, implementation, test, and deploy). In aspecific project, recursion and repetition of phases areallowed. From the macro level, however, these phases canbe regarded as generic for most business solutions. On thebasis of this common lifecycle, Model Blue methodologycan easily be mapped and integrated using existingsolution methodologies and software engineering approach(e.g., Rational Unified Processes). In this specific case,Model Blue is integrated with the e-business Customengagement model in the IBM IGS Method, which is awidely adopted services methodology within IBM. Themodels built by practicing Model Blue (and documentsgenerated) are successfully integrated as a part of (or areplacement for) the work-product deliverables definedin the IBM IGS Method. Since these models are builtin collaboration with Bank SinoPac as part of thespecification and design process, the semi-auto-generatedwork product (requirements specification, macro design,micro design, test-case specification, etc.) is easilyaccepted by the corresponding owners of the receivingorganizations.

In the remainder of this section, we describe themethodology in detail, including the major models created,transformed, updated, and utilized in each phase. Keyexperiences received from the engagement are alsohighlighted.

Figure 9

Model Blue engagement lifecycle and output.

Requirement Macro design Micro design Implementation Test

Model Blue

(Business View)

and Information

Artifact template

Generated

document

(HTML)

Requirement

document (DOC)

Model Blue

skeleton

(IT View)

Generated

document (HTML)

Component

relation diagram

Design

document (DOC) UML

models

XML

schema/DTD

WBI

Collaboration

template

Model Blue

(IT View)

WBI BO

definition

Test case

definition

Java source

code

Model Blue

(IT View)

simulated

Deploy

Model Blue

engine

WBI

server

Customer

WBI

Collaboration

Transformation/generation

Revise/guide/deploy


659

Requirement specificationIn the requirement specification phase, businessconsultants interview the Bank SinoPac staff to writedown their “as-is” and “to-be” business process models.The objective is to discuss, capture, and document theinternal process logic as well as interprocess interactionrelationships clearly, so that all related parties can fullyunderstand the business process requirements.

Deliverables

● Business View models Eclipse-based tooling is used byconsultants to create the two-dimensional Business Viewmodels. The tool provides an integrated modelingenvironment with capabilities of diagram drawing,importing and exporting, assembling, verifying, printing,etc. Some add-on capabilities, such as detecting conflict

among assembled business processes, are also provided.The generated model is an XML file with both syntactic(XML schema) and semantic (well-formed rules withtooling support) correctness definitions.

● Information Artifact templates The Information Artifact(IA) template is the meta-level definition of dataelements that are exchanged across different businessprocesses and entities. One example of an IA templateis the account transfer request and response messagesthat are exchanged with the core banking system. Thedata structure is defined in a tree-view editor andsaved as an XML file, which is then bound with theInformation Artifact construct in the Business Viewmodel, as illustrated in Figure 10.

● HTML format requirement documents HTML formatrequirement documents are automatically generatedfrom the above models as a recordable representation

Figure 10

Business View builder: Information Artifact definition.


660

of the process description. They also contain informationregarding many nonvisualized yet valuable inputs duringthe interview (e.g., some business rules definitions andtext-formatted comments). The HTML documents arethen merged into the requirement document.

Key experiences gained

● For the Bank SinoPac staff, an intuitive and simplemodeling method with tooling to document or recordtheir requirements (focused on data and processelements) is a major advantage.

● One challenge which remained unsolved was how toeffectively manage and handle all of the businessterminologies used in the models and design documents.The introduction of ontology or vertical standards toregulate the vocabulary usage seems to be the solution,

yet most existing ontology specifications today (i.e.,Resource Description Framework [34, 35] and WebOntology Language [36]) are too heavyweight to beput into practical usage.

Macro designIn the macro design phase, business consultants worktogether with solution architects to produce an overallarchitecture of the target system. The objectives of thisphase include designing the execution process andidentifying the high-level component architecture.

Deliverables

● IT View models (skeleton) An initial IT View model isautomatically generated from the Business View model.The flow execution logic embedded in the original

Figure 11

IT View designer: Model Blue IT View model.


661

diagram is analyzed, extracted, and then transformedinto an equivalent tree-structured representation as thestarting point for IT execution flow design. Furtherprocess design then follows (e.g., adding some actionnodes to describe implementation details such as loggingand data transformation, which are not the majorconcern for business consultants). The Business Viewdesigner, transformation engine, and IT View designerare seamlessly integrated to provide unified end-to-enduser experience.

● Component Relationship Diagram (CRD) A high-levelarchitecture for all key function components within thesystem is created in the macro design phase. Typicalcomponents may include the integration hub, back-endsystems, database and directory server, and utilityservices. A component relationship diagram is thencreated to show all of the components as well as theinteraction relationships among them. There is no needto produce a detailed design for each component in thisphase.

● Mapping from IT View model to components Mappingsfrom the IT View model to related components areidentified in this phase by introducing new componentsif they are not yet included in the CRD. For example, ifone action node in the IT View stands for the invocationof a performance logging that is not included in theCRD, a new performance logging component isadded and a mapping is established from the actionnode to the component.

● Adapters/Collaboration list For an integration project,one important element is the connection to variouslegacy systems. In the macro design phase, all keyadapters required are identified. For this specificengagement, the application connection layer isimplemented with IBM WebSphere Business Integration(WBI), which provides many standard connectors tolegacy systems (e.g., IBM WebSphere MQ Series,Unisys, and ecSolution).


● With a few days of training and practicing, the IT Viewmodel and tools can be understood and used by the ITstaff. However, the business staff needs more time tounderstand the structured representation of processflows.

● With IT View models, reusable components are easilyidentified and organized, and as a result thedevelopment phase is greatly accelerated.

Micro designIn the micro design phase, solution architects and processdesigners continue their work on the macro design to

make it complete and executable. The objective of thisphase is to create a design specification with enoughinformation for programmers to develop the solution.

Deliverables

● Model Blue IT View models (simulated version) With thehelp of the Eclipse-based IT View designer (Figure 11),initial versions of IT View models are elaborated in themicro design phase toward executable scripts, whichcan then be deployed on the engine to drive businessexecution. Simulation support is provided for the ModelBlue IT View model (Figure 11) so that one can verifythe correctness of the flow structure before real codingis started. The simulation engine is built with the samekernel of the Model Blue IT View Engine that actuallyexecutes the IT View model. The major difference liesin that input and output from legacy systems areredirected to local file systems so that one can easilyemulate back-end systems to produce various scenarios.

● Finalized adapter (Collaboration) and action list On thebasis of the results of the finalized version of the ITView models, a table is generated to list all of theadapters (WBI Collaboration) and service actions thathave to be implemented. It will be used as a basis tocreate job specifications for contract programmers andtesters.

● Information schema specification A transformationengine is provided to automatically generate astandard XML schema definition from the associatedInformation Artifact template definition in the BusinessView model. The XML schema is then used as thestandard data schema specification for the wholeproject team.

● WBI 2 Collaboration template and Business Object (BO)definition For this project, one specific requirementfrom Bank SinoPac was to fully leverage existing WBIconnectors; therefore, some parts of the functionsrelated to back-end system connection (usually in theform of a subtree of the IT View) are implemented inthe form of WBI Collaboration. At runtime, the ModelBlue IT View Engine is responsible for executing mostflow control logic (such as branch and parallel) andinvoking certain functions on the WBI server to accessback-end mainframes. To drive the implementationphase smoothly, we developed a transformation engine

2 WebSphere Business Integration (WBI) is a collection of IBM software productsfor business integration which includes the following core capabilities: Model,Integrate, Connect, and Monitor. Because of issues related to customeracceptance and product lifecycle, two major components of WBI were used in thisengagement: WebSphere Business Integration Collaboration, which provides pre-built process templates for common business practices, and WebSphere BusinessIntegration Server, which provides a runtime to execute WBI collaboration.


662

to automatically turn a subtree of the Model Blue ITView into an equivalent representation of the WBICollaboration template.

For the information specification part, thegenerated XML schema is transformed into XMLdata type definition (DTD) and then converted byWBI tools into the definition of the Business Objectin WBI.

● Detailed component specification On the basis of thecomponent relation diagram, detailed designs in theform of UML models (i.e., class diagram, sequencediagram, and state diagram) are created for thosecomponents that must be created during theimplementation phase.

● HTML format design documentation As in therequirement-collecting phase, HTML format designdocuments are automatically generated from ModelBlue IT View models to create a recordablerepresentation of the whole system design.


● Since it is an interface common to different sections ofthe system, the information schema definition proves tobe the critical point in project lifecycle management.

● The Eclipse-based tooling environment is welcomed bybusiness consultants, IT architects, process designers,and programmers. The integrated tooling environmentand workspace increase the efficiency of knowledgesharing and discussion among them.

ImplementationIn the implementation phase, the software code for thesystem is developed. The objective of this phase is tocreate the development document, software code, andtest cases.

Deliverables

● Development task guide document To let developersunderstand clearly the specific tasks assigned to them,a document is provided for each task in the developerbase of the Model Blue IT View model. This documentincludes the guide for the program interface, majorprocessing logic, input/output parameters, Java sourcecode skeleton (all of the above items generated fromthe Model Blue IT View model), skills needed,etc.

● Test cases In this phase, test cases for the test phase areprepared. The IT View models are analyzed to createthe testing scripts for business process models. The treetraversal structure helps clarify all possible processexecution scenarios. As a result, it reduces the timeneeded to prepare the test cases and enhances thecoverage of them.


● XML is universally used for the data exchange fromfront end to integration hub and then to back-endconnectors. One problem is the performance overheadcaused by XML tree parsing. We recommend that futureintegration solutions include native XML support (forexample, including XML as an internal data type so thatunnecessary transformations between strings and theXML tree can be avoided).

● Outsourcing control is a major issue in a relativelylarge-scale development project. We have combinedmultiple models into a development task guidedocument; the result is very encouraging in that mostcontractors can understand their jobs as described bythe document quite well, while the document containsno additional information that they do not need.

TestIn the test phase, architects, testers, developers, and BankSinoPac staff work together to verify the target systemagainst its requirements. The test phase is actually dividedinto four stages: Unit Test, Function Verification Test(FVT), System Integration Test (SIT), and UserAcceptance Test (UAT).

Deliverables

● Testing document A summary of the testing work ispresented in the delivered documents. It includes thepurpose and principle for testing, a description of eachtest case (scenario, input, expected output, owner, etc.),results for each test case, and a description of eachvariance (severity, scenario, input, output, owner,handling result, etc.).

● Performance test report Performance is a major concernof a production system. In this project, we used theRational Team Test suite and Websphere ResourceAnalyzer as integrated tooling for the performance test.Two major system performance factors, throughput andresponse time, were generated under different testingscenarios (e.g., different concurrent users).


● Testing of an integration project does involve multipleback-end systems. In the FVT stage, when no back-end connectors were available, we had to buildadditional programs to emulate different back-end mainframe transactions. Even in the SIT stage,when we already had the real adapter code frompartner teams, it was still difficult to actually connect toback-end mainframes that were running mission-criticalapplications.


663

● IT View provides an integrated visual debuggingenvironment at build time. Integrated debugging canhelp to detect logic defects between flow and actionsbefore the deployment. This shortened the test andrevision cycle and reduced the testing time significantly.

● When performing FVT and SIT of process flows, simplyverifying the return XML message is not enough. Thesystem logs of the engine should be inspected to obtaina detailed execution history.

● According to the performance test, XML processingis a memory- and CPU-intensive task. The techniqueof XSLTC [37] can boost the efficiency of XMLtransformation. One approximate test (on a PC)revealed that introducing XSLTC with other settingsunchanged reduced processing time by at least 60%.

Lifecycle summaryIn summary, models are used as the key element to drivethe whole BPIM solution. A set of models is taken asinput reflecting each phase of the lifecycle. With definedsteps of model creation and revision, another set ofmodels (also including the generated documents frommodels) is generated as output. Model Blue is morefocused on the business process model, whether it is in theBusiness View or IT View, so the model-driven approachcan be labeled as process-centric. The process-centricapproach fits well with Bank SinoPac�s integrationpurpose—to support the introduction of new financialservices and products by creating business processes toconnect various back-end systems, workers, and resources.We recognized the fact that some particular integrationscenarios require modeling to be focused on other aspects

[e.g., data modeling supported by the IBM DB2Information Integrator (DB2 II)]. However, the keychallenge for these solutions lies in effectively organizingthe data view around the usage context, which is usuallythe business processes that consume these data.

4. Best practices and lessons learned

Model transformationModel transformation is critical for the smooth executionof a model-driven approach. It is regarded by some as theheart and soul of model-driven software development[38]. In Model Blue, we have developed differenttransformation engines serving different purposes, asillustrated in Figure 12:

● The “Business 3 IT” transformation engine, althoughnot full-fledged, has demonstrated its capability as abridge between the business specification and ITimplementation models. We can fill the so-calledbusiness–IT gap even better with more transformationcapabilities (e.g., how to do round-trip transformationfrom the IT View to the Business View, and how totransform the data mapping relationship within theBusiness View into corresponding XSLTC scripts). Themost challenging problem, however, lies in propagatingchanges from the Business View to the IT View whensome later changes are required.

● The “Information Artifact template 3 XML schema 3WBI BO” transformation engine is a key step for thehandover of data models from a business person�sperspective to standard IT representation.

● The “IT View 3 WBI Collaboration template”transformation engine is intended to simplify the workof collaboration designers with a model transformationapproach, yet the results have not been ideal. Thecollaboration designers were reluctant to learn a newmodeling paradigm that did not greatly reduce theirworkload. This was partly because most collaborationused by this engagement is relatively simple and thuscan be easily hand-coded. The value of the IT Viewmodel was its capability to define and execute (with theModel Blue IT View Engine) complex process logic(e.g., parallel execution) that was not supported byWBI Collaboration.

In summary, model transformation has played animportant role in connections for the whole projectlifecycle. However, we should never overestimate thepower of model transformation and overlook thecomplexities it may introduce. One guideline for modeltransformation is to use this approach only when it reallybrings value or cannot be replaced by other efficientalternatives.

Model transformation in Model Blue.

Figure 12

Model Blue

Business View

Flow execution

structure

Major change

Action binding

Inter-flow call

Performance optimization

Condition/task

Logging

Performance

Exception

Information

Artifact

Traceability

XML

schema

Model Blue

IT View

(skeleton)

Model Blue

IT View (full)

WBI

Collaboration

template

Subtree

WBI tool

WBI BO


664

Asset managementWe tried three different methods of managing the vastnumbers of assets (models and documents) across theproject lifecycle. The first was to use a Lotus Notes*database as a shared repository for model and documents.Outside resources, such as supplemental and contractpersonnel, needed Notes IDs to access the database.We also had difficulties in synchronizing updates.Notes is currently not well integrated with developmentenvironments such as Eclipse and WebSphere ApplicationDeveloper (WSAD).

The second choice for us was to use the open-sourceConcurrent Version System (CVS) [39] to share all relatedmodels and code. CVS provides a lightweight teamdocument sharing tool which is seamlessly integrated withWSAD and Eclipse. For this project of moderate size (nomore than 50 developers and testers), we chose CVS asthe collaboration server, mainly because of the efficiencyboost.

The third alternative was to use the Model Blue AssetManager, which is designed to support asset sharing andreuse in an enterprise environment. Asset Manager hadall of the capabilities required by a large-scale projectenvironment, including complex categorization and accesscontrol capabilities (integration with enterprise directoryserver); however, it was heavy-weight (requiring WAS,DB2, etc.) and thus hard to install and configure.

The adoption of the CVS server illustrated the fact thatmost people wish to have easy-to-use, lightweight systems.Packing too many functions into one solution may, contraryto initial expectations, greatly increase the complexity andexclude the solution from practical consideration.

Change managementIn an integration project involving many enterprisebusiness applications and processes, requirement changeswill inevitably have a significant impact on the wholeproject. Discipline with respect to change estimates andapprovals thus becomes very important. During theproject, the IBM project team experienced seven majorrequirement changes that required Bank SinoPac to signoff officially, in addition to numerous other minor changesthat were contained internally by the project team. TheModel Blue Business View provided a clear view of therequirement change at the business operation layer. Thiswas used to illustrate and discuss the change requirementswith Bank SinoPac to better understand what really hadto be changed. The clear mapping relationship from theBusiness View to the IT View made it possible for theIBM project team to demonstrate various effects of thechanges proposed, and at the same time limit the changepropagation to a controllable scope. The Model Blue ITView provided a base for calculating additional resource

and time needed to make a change possible. For example,when we had to change the data format of a specificbusiness object, we went through the IT View models tofind all actions that utilize the data and thus have to berevised accordingly, and then determined how muchworkload (in terms of person-days) was required andhow much delay was expected.

The ideal method of model-driven change managementwould be to propagate changes automatically from theBusiness View to the IT View without major manual work.For Model Blue, this capability is not yet supported, partlybecause it is still unrealistic for a transformation engine toconvert everything from the Business View into the ITView. The support of Model Blue on change managementis still focused on the facilitation of requirementillustration, solution discussion, and impact estimationfor customer change requests. Even so, the model-drivenapproach has demonstrated its potential for solving theproblem of scope and change control, which are essentialfactors for the on-time and on-budget delivery ofintegration projects.

Skill transferTransferring skills to the Bank SinoPac staff and enablingthem to use and operate the Model Blue methodology waschallenging, given that most of them were new to evenXML and Java. The IBM team used various alternativesto help them gain the ability to use the tool andmethodology. The China Research Laboratory (CRL)hosted two face-to-face workshops with the Bank SinoPacbusiness and IT staff and business consultants in BCS.CRL also delivered a week of lecture-based training at theBank SinoPac site for about 30 Bank SinoPac staff fromdifferent departments. CRL and BCS delivered a week ofon-site training in Taipei after the lecture, working withfive Bank SinoPac employees from the IT department ona sample integration scenario. Yet another approach usedby the BCS team was to invite Bank SinoPac to send theiremployees to work with them. A BCS project managerthen assigned work to the Bank SinoPac employees as ifthey were IBM team members. Through several months ofworking together, the Bank SinoPac employees becamefamiliar with most of the details about the projectdevelopment lifecycle. The skill transfer results werevery good.

Since the IT experience, knowledge, and capabilitiesof a customer�s staff are difficult to predict in mostengagements, deploying a new methodology can imposemajor risks in execution and eventual acceptance. In thisproject, we believe that the lightweight nature of ModelBlue played an important role in making it much moreaccessible and acceptable to new users.


665

5. SummaryA set of model-driven business integration andmanagement methods, frameworks, supporting tools, andruntime environment was developed and demonstrated.An entire solution was provided, from Business ProcessModeling at the operations level, through the executionand implementation levels, all the way to coding andtesting. The Model Blue Business View provided a two-dimensional modeling capability that mixed flow modelingand data modeling. This, plus its Eclipse-based tooling,made it a powerful tool for Bank SinoPac to use indefining their business requirements with good structureand detail. Without this, the project team would not havebeen able to collect requirements and specifications insuch a short time. The Model Blue IT View provided atree-based model to help the Bank SinoPac IT staff workwith the IBM project team to identify their IT goals,limitations, and expectations. The project team wasalso able to explore a number of different designs andimplementation approaches before coding. The ModelBlue IT View helped the geographically distributed projectteam to understand the complete working environment fortheir job assignment. It helped to allocate work items tooutside contract programmers and testers in terms ofjob assignment specification, job context, and qualityassurance criteria. The Model Blue IT View also helpedto generate testing scripts for the business process,dramatically reducing the time needed to prepare the testcases and enhance test-case coverage. The Model Blue ITView Engine provided a very lightweight WAS-basedscript execution environment to run the customer�sbusiness processes by routing through various businessapplications and/or transactions. The combination ofModel Blue runtime and the WBI Server achieved thecustomer�s goal of high flexibility and high performance.

BPIM methods and models were seamlessly integratedin Model Blue with existing development methodologies,with full lifecycle tooling support. The models anddocuments built by practicing Model Blue weresuccessfully integrated as a part of (or a replacementfor) the work-product deliverables defined in the IBMGlobal Service Method. Since these models were built incollaboration with the customer as part of the specificationand design process, the customer was very happy withthe semi-auto-generated work product (requirementsspecification, macro design, micro design, test-casespecification, etc.). Compared with approaches employedon similar projects, the model-driven approach was up to30% more efficient. The work was done on time, withinbudget, and with extremely high customer satisfaction,despite remote locations and environmental restrictionswhich prevented the team from having face-to-facemeetings and discussions.

Changes in requirements were contained because ofthe clear modeling layers in Model Blue and the clearmapping among them. The project team was able todemonstrate various effects for the changes proposed bythe customer, in terms of both customer function andimplementation impact. Throughout the project, therewere seven major requirement changes after designspecifications were officially signed off by the customer, aswell as numerous other “small or incremental” changes.

6. Opportunities and challengesDespite all of the positive feedback received from users,the methods and tools described in this paper are far fromthe degree of perfection which model-driven technologiescan achieve. IBM and Bank SinoPac are planning futurecollaborations to further explore the opportunities andchallenges of using a model-driven approach to addressmore complex business and IT problems. Future topicsinclude simple, flexible, and configurable business processmodeling capability; modeling additional existing businessprocesses for Bank SinoPac; strengthening the theoreticalfoundation to include more analytic features; exploringopportunities for research on business semantics; andapplying the model-driven business integration approachto industries other than banking.

AcknowledgmentsThe authors are grateful to the management teams ofBank SinoPac, BCS and CRL, especially Joan N. Fang,S. B. Hsieh, Cliff Yu, Max Kuo, James Yeh, and LouZhou, for their persistent support for the project acrossthe whole project lifecycle. Some of the initial Model Blueideas came from early work (i.e., Information FunctionFlow) from the IBM Thomas J. Watson Research Center,and here we would like to express our appreciation to Jen-Yao Chung, Kumar Bhaskaran, Santhosh Kumaran, YingHuang, Zongwei Luo, Nathan Caswell, and Nigam Anil.We owe thanks to other project participants for theirideas and efforts, including Denise Chen, Wei-Chih Yu,Andy Chen, Winston Huang, Ching-Ting Pan, Nan-NiHsu, and Annie Chen from Bank SinoPac; Hank Chenand Jesse Ko from IBM BCS; and Xin Zhang, GuanqunZhang, Jing Li, Jian Wang, Haiqi Liang, Tony J. Liu, andWalter Yang from IBM CRL. The authors also appreciatethe assistance from Dong Liu and Xiulan Yu in reviewingthis paper. Finally, special thanks should go to HelenZhang for her great contributions to the earlier versionof Model Blue.

*Trademark or registered trademark of International BusinessMachines Corporation.

**Trademark or registered trademark of MicrosoftCorporation, BEA Systems, Inc., iGrafx (a division of Corel,Inc.), or SmartDraw.com.


666

References1. Hurwitz Group, “Connecting the Enterprise: iWay

Software”; see http://www.iwaysoftware.com/products/whitepapers.html.

2. IBM Corporation, IBM Websphere Business Integrationhome page; see http://www.ibm.com/software/info1/websphere/index.jsp?tab�products/businessint.

3. Microsoft Corporation, Microsoft BizTalk Server homepage; see http://www.microsoft.com/biztalk/.

4. BEA Systems, Inc., BEA WebLogic Integration homepage; see http://www.bea.com/products/weblogic/integration/index.shtml.

5. N. G. Carr, “IT Doesn�t Matter,” Harvard Business Rev.,pp. 5–12 (May 2003).

6. K. Nelson and H. Nelson, “The Need for a StrategicOntology,” Proceedings of the MIS (ManagementInformation Systems) Quarterly Special Issue Workshopon Standard Making: A Critical Research Frontier forInformation Systems, Seattle, December 12-14, 2003, pp.361–371.

7. Object Management Group, Model Driven Architecture;see http://www.omg.org/mda/.

8. IBM Corporation, IBM WebSphere MQ Workflow homepage; see http://www.ibm.com/software/integration/wmqwf/.

9. Business Process Execution Language for Web ServiceSpecification; see http://www.ibm.com/developerworks/webservices/library/ws-bpel/.

10. Eclipse.org, Eclipse White Paper; see http://www.eclipse.org/whitepapers/eclipse-overview.pdf.

11. H. Liang, J. Wang, Z. Tian, L. Zhang, J. Zhu, and W.Ding, “Managing Business Process Models: A PrivateUDDI Application Study,” Proceedings of the 6th WorldMulti Conference on Systemics, Cybernetics, and Informatics(SCI 2002), pp. 408 – 412.

12. T. Murata, “Petri Nets: Properties, Analysis andApplications,” Proc. IEEE 77, No. 4, 541–580 (1989).

13. W. M. P. van der Aalst, “Application and Theory of PetriNets,” in Verification of Workflow Nets, P. Azema and G.Balbo, Eds., 1997; published in Lecture Notes in ComputerScience 1248, 407– 426 (1997).

14. W. Ding, Z. Tian, L. Zhang, J. Wang, J. Zhu, and H.Liang, “Conflicts Analysis for Inter-Enterprise BusinessProcess Model,” Proceedings of the 6th World MultiConference on Systemics, Cybernetics, and Informatics (SCI2002), pp. 245–250.

15. D. Georgakopoulos, M. Hornick, and A. Sheth, “AnOverview of Workflow Management: From ProcessModeling to Workflow Automation Infrastructure,”J. Dist. & Parallel Databases 3, No. 2, 119 –153 (1995).

16. W. M. P. van der Aalst and A. H. M. ter Hofstede,“Workflow Patterns: On the Expressive Power of (Petri-Net-Based) Workflow Languages,” Proceedings of theFourth Workshop on the Practical Use of Coloured PetriNets and CPN Tools (CPN 2002), Vol. 560 of DAIMI, pp.1–120; University of Aarhus, Aarhus, Denmark, August2002.

17. The Banker online; see http://www.thebanker.com/news/fullstory.php/aid/564/Taiwan.html.

18. Euromoney online; see http://www.euromoney.com/awards2003/.

19. Global Finance online; see http://globalf.vwh.net/content/?article_id�485/.

20. Enhanced Messaging and Queuing home page; see http://sybooks.sybase.com/onlinebooks/group-emq/eqg0260e/.

21. J. A. Whittaker and H. H. Thompson, “Black BoxDebugging,” ACM Queue (issue on DistributedDevelopment) 1, No. 9, 68 –74 (December/January 2003/2004).

22. Object Management Group, Unified Modeling Language;see http://www.omg.org/uml/.

23. IBM Corporation, IBM Rational Rose; see http://www.rational.com/.

24. iGrafx FlowChart home page; see http://www.igrafx.com/products/flowcharter/.

25. SmartDraw home page; see http://www.smartdraw.com/.26. J. A. Zachman, “A Framework for Information Systems

Architecture,” IBM Syst. J. 26, No. 3, 276 –292 (1987).27. Popkin Enterprise Architecture; see http://

www.popkin.com/products/system_architect/enterprise_architecture.htm.

28. J. D. Mooney, “Portability and Reusability: CommonIssues and Differences,” Proceedings of the ACM 23rdAnnual Conference on Computer Science, Nashville, TN,1995, pp. 150 –156.

29. W. Frakes and C. Terry, “Software Reuse: Metrics andModels,” ACM Computing Surv. 28, No. 2, 415– 435(1996).

30. M. Turnlund, “Distributed Development LessonsLearned,” ACM Queue (issue on DistributedDevelopment) 1, No. 9, 26 –31 (December/January 2003/2004).

31. K. Coar, “The Sun Never Sets on DistributedDevelopment,” Ibid., pp. 32–39.

32. G. Joeris, “Change Management Needs Integrated Processand Configuration Management,” Proceedings of the 6thEuropean Conference, held jointly with the 5th ACMSIGSOFT International Symposium on Foundations ofSoftware Engineering, November 1997, Vol. 22, No. 6, pp.125–141.

33. IBM Corporation, IBM Rational Unified Process homepage; see http://www-3.ibm.com/software/awdtools/rup/.

34. RDF (Resource Definition Framework) Primer, W3CProposed Recommendation, December 2003; see http://www.w3.org/TR/2003/PR-rdf-primer-20031215/.

35. S. Decker, P. Mitra, and S. Melnik, “Framework for theSemantic Web: An RDF Tutorial,” IEEE InternetComputing 4, No. 6, 68 –73 (November/December 2003).

36. OWL Web Ontology Language Overview, W3C ProposedRecommendation, December 2003; see http://www.w3.org/TR/2003/PR-owl-features-20031215/.

37. XSLTC home page; see http://xml.apache.org/xalan-j/xsltc/.38. S. Sendall and W. Kozaczynski, “Model Transformation:

The Heart and Soul of Model-Driven SoftwareDevelopment,” IEEE Software 20, No. 5, 42– 45(September/October 2003).

39. Concurrent Versions System home page; see http://www.cvshome.org/.

Received December 31, 2003; accepted for publication


667

February 25, 2004; Internet publication September 16, 2004

Jun Zhu IBM Research Division, IBM China ResearchLaboratory, HaoHai Building, No. 7, 5th Street, Shangdi,Haidian District, Beijing 100085, People�s Republic of China([email protected]). Mr. Zhu is a Research Staff Memberon the Business Integration team at the IBM China ResearchLaboratory (CRL). He received his B.S. and M.S. degreesin computer science from Shanghai JiaoTong Universityin Shanghai, China, in 1998 and 2001, respectively. Hesubsequently joined the IBM Research Division in Beijing,China, where he began working on dynamic e-businesstechnologies and enabling systems. Since 2002, Mr. Zhuhas focused his attention on business process modeling andintegration technologies, in particular the applicability ofmodel-driven technologies to solve business integrationproblems. He contributed significantly to the concepts anddesign of the model-driven business integration platform,Model Blue, specifically Business View modeling and end-to-end solution methodology. Mr. Zhu was also one of theleaders of the CRL technical team to engage Model Bluewith a customer—Bank SinoPac—in 2003.

Zhong Tian IBM Research Division, IBM China ResearchLaboratory, HaoHai Building, No. 7, 5th Street, Shangdi,Haidian District, Beijing 100085, People�s Republic of China([email protected]). Dr. Tian joined the IBM China ResearchLaboratory after receiving his Ph.D. degree from FudanUniversity, Shanghai, China, in 1995. Since then, he hasworked in a number of areas, including Digital Library, B2BeCommerce, e-Procurement, dynamic e-Business, and model-driven business integration. He was among the first pioneersto advocate Digital Library in China. He was one of the keyResearch architects of the WCS Marketplace edition. Byworking with IBM Global Procurement, Dr. Tian and histeam helped transform two major IBM production sitesinto e-Businesses. Most recently, he led a joint team fromResearch and BCS to successfully deliver a model-drivenbusiness integration platform for Bank SinoPac.

Thomas Li IBM Research Division, IBM China ResearchLaboratory, HaoHai Building, No. 7, 5th Street, Shangdi,Haidian District, Beijing 100085, People�s Republic of China([email protected]). Dr. Li is the Deputy Director andCTO of the IBM China Research Laboratory. He holds a B.S.degree in electrical engineering, an M.S. degree in computerscience, and a Ph.D. degree in management informationsystems. His business-oriented knowledge and experience havebeen instrumental in formulating the direction and strategy ofnumerous organizations in both Asia and the U.S. Dr. Li hasmore than twenty years of experience in the computer andInternet industry. During the past five years, he has been oneof the major drivers in several Internet initiatives in the U.S.and Asia. These multinational projects include the areas ofproactive and personalized customer services, a multiregionalon-line shopping site, and intelligent multichannel large-volume information delivery. His technical expertise andinnovative thinking have led to 39 patents in objecttechnology, digital communication, visualization tools, anddatabase management systems, and he is currently workingon better managing and modeling strategic business change.In addition to publications in journals, proceedings, andtechnical bulletins, he has been one of the key contributors indelivering eight software commercial products, three hardwaresystems, and a number of architectural designs and technicalspecifications.

Wei Sun IBM Research Division, IBM China ResearchLaboratory, HaoHai Building, No. 7, 5th Street, Shangdi,Haidian District, Beijing 100085, People�s Republic of China([email protected]). Dr. Sun is a Research Staff Memberon the Business Integration team at the IBM China ResearchLaboratory (CRL). He received his Ph.D. degree in electronicengineering from the Beijing Institute of Technology in 2001.He subsequently joined the IBM Research Division in Beijing,China, where he began working on B2B integration platformtechnologies. Data extraction and transformation, processinglogic runtime, and design time support have been his researchfocus. Dr. Sun contributed significantly to the concepts anddesign of Model Blue with respect to the IT View modelingand execution framework. He was also a leader of the CRLtechnical team to engage Model Blue with a customer—BankSinoPac—in 2003.

Sheng Ye IBM Research Division, IBM China ResearchLaboratory, HaoHai Building, No. 7, 5th Street, Shangdi,Haidian District, Beijing 100085, People�s Republic of China([email protected]). Mr. Ye is a Research Staff Memberon the Business Integration Methodology team at the IBMChina Research Laboratory (CRL). He received his B.S.and M.S. degrees in computer science from Xi�an JiaoTongUniversity in Xi�an, China in 1997 and 2000, respectively. Hesubsequently joined the IBM Research Division in Beijing,China, where he began working on e-procurement and dataformat transformation technologies. Since 2002, Mr. Ye hasfocused on lightweight business process runtime engines andtooling. He was one of the major designers of Model Bluefor the IT View modeling, including the runtime engine andtooling. Mr. Ye was the major contributor to the integrateddebugging environment for the Model Blue IT View. He wasalso a member of the key CRL technical team to engageModel Blue with a customer—Bank SinoPac—in 2003.

Wei Ding IBM Research Division, IBM China ResearchLaboratory, HaoHai Building, No. 7, 5th Street, Shangdi,Haidian District, Beijing 100085, People�s Republic of China([email protected]). Dr. Ding is a Research Staff Member inthe On Demand Technologies Department at the IBM ChinaResearch Laboratory (CRL). She received her Ph.D. degreein management science from Harbin Institute of Technologyin 2001, subsequently joining the IBM Research Divisionin Beijing, China, where she began working on dynamice-business technologies and enabling systems. From 2001to 2002, Dr. Ding focused her attention on business processmodeling and integration technologies, and in particular, theapplicability of model-driven technologies to solve businessintegration problems. She was the major designer for aconflict-detection algorithm based on formalized mathematicalfoundations, which is a key function for Model Blue. Dr. Dingwas a member of the key CRL technical team to engageModel Blue with a customer—Bank SinoPac—in 2003.

C. C. Wang Information Technology Division, BankSinoPac, 12F, 66, Lane 9, Shung Shan Road, Taipei 105,Taiwan ([email protected]). Mr. Wang is a VicePresident in charge of overall e-commerce software systemdevelopment at Bank SinoPac. He received his B.S. degree incomputer science from Tamkang University in Taipei, Taiwan,in 1985. He maintained and developed a banking mainframe(Unisys) online system for the Land Bank of Taiwan from1986 to 1991. In 1991 he joined the Information Technology


668

Division of Bank SinoPac in Taipei, Taiwan, where he beganworking on administering and planning a banking mainframe(Unisys) system; he was also the leader of the bankingmainframe software team responsible for online systemdesign, development, and maintenance. Since 2000, Mr.Wang�s assignment has been related e-commerce softwaresystems, in particular, the applicability of Internettechnologies to fulfill the business requirements of the BankSinoPac Corporate Banking Division and Retail BankingDivision. Mr. Wang managed all of the e-commerce softwareteams in the Information Technology Division of BankSinoPac. These teams are responsible for the design,development, and maintenance of an EAI system, a B2BWeb site, and a B2C Web site. He was also the BankSinoPac project manager for the CRL/Bank SinoPacModel Blue engagement in 2003.

Gerland Wu Information Technology Division, BankSinoPac, 12F, 66, Lane 9, Shung Shan Road, Taipei 105,Taiwan ([email protected]). Mr. Wu is an Assistant VicePresident in charge of EAI software system development atBank SinoPac. He received his M.S. degree from NationalTsing Hua University, Taiwan, in 1990 and his B.S. degreefrom Fu Jen University, Taiwan, in 1988, both in appliedmathematics. He subsequently worked for the Institute forInformation Industry in Taiwan from 1990 to 1998 as asoftware programmer in a financial team and participated inthe development of the Transactional Processing Monitormiddleware of OnLine Transaction Processing (OLTP)on a UNIX platform. In 1999 he joined the InformationTechnology Division of Bank SinoPac in Taipei, Taiwan,where he began work maintaining Internet banking and WAPmobile banking. Since 2000, Mr. Wu�s assignment has beenB2C-related e-commerce system development—in particular,providing EAI middleware to fulfill the Bank SinoPace-commerce requirements. Mr. Wu leads the EAI applicationteam in the Information Technology Division of BankSinoPac. His team is responsible for the design, development,and maintenance of EAI middleware to support MMAB2Band MMAB2C Web sites and integrate back-end legacysystems. In addition, he was one of the members whoproposed the technical requirements for the CRL/BankSinoPac Model Blue engagement in 2003.

Leo Weng Information Technology Division, Bank SinoPac,10F, 66, Lane 9, Shung Shan Road, Taipei 105, Taiwan([email protected]). Mr. Weng is an Assistant VicePresident in charge of B2B e-commerce software systemdevelopment at Bank SinoPac. He received his M.B.A. degreein management information systems from Simon FraserUniversity in Vancouver, Canada, in 1996. From 1992 to 1994he worked for Unisys Taiwan as a software programmer. Hejoined the Information Technology Division of Bank SinoPacin Taipei, Taiwan, in 1996 to work on maintaining a bankingmainframe batch system. Since 1998, Mr. Weng�s assignmenthas been B2B-related e-commerce system development—inparticular, applying Internet technology to fulfill the businessrequirements of the Bank SinoPac Corporate BankingDivision. Mr. Weng was the leader of the B2B applicationteam in the Information Technology Division of BankSinoPac. His team is responsible for the design, development,and maintenance of the Bank SinoPac MMAB2B Web site.In addition, he also acted as the Bank SinoPac corporatecustomers� Internet portal for the end product of theCRL/Bank SinoPac Model Blue engagement in 2003.

Simon Huang Application Innovation, Business ConsultingServices, 3F, No. 7, Song-Ren Road, Taipei 110, Taiwan([email protected]). Mr. Huang is the EAI solutionoffering manager of IBM Taiwan. He received his B.S. degreein computer science from Soochow University in 1988 and hisM.S. degree in computer science from National Tsing-HuaUniversity in 1990. He joined IBM Taiwan in 1996 and hasfocused on EAI and workflow integration solutions since1998. During these years, he led the team which successfullyimplemented more than ten EAI/workflow-related projectsin Taiwan, including the Bank SinoPac engagement.

Bill Liu Application Innovation, Business ConsultingServices, 3F, No. 7, Song-Ren Road, Taipei 110, Taiwan([email protected]). Mr. Liu is the Project Manager of IBMTaiwan, and is a certified Project Management Professionalof the Project Management Institute. He joined IBM Taiwanin 1997 and has focused on e-business solutions since 1998.During these years, he led the team which successfullyimplemented several e-business projects in Taiwan,including the Bank SinoPac engagement.

Daniel Chou Application Innovation, Business ConsultingServices, 3F, No. 7, Song-Ren Road, Taipei 110, Taiwan([email protected]). Mr. Chou is the EAI solutionarchitect of IBM Taiwan. He received his B.S. and M.S.degrees in computer science from National ChiaotungUniversity in 1993 and 1995, respectively. He joined IBMTaiwan in 1999 and has focused on EAI and workflowintegration solutions since 2001. During these years, hedesigned and successfully implemented more than fiveEAI/workflow-related projects in Taiwan, including theBank SinoPac engagement.


669

model-driven business process integration and management: a … · 2015-07-28 · and...

Documents