This tool presentation gives insights into the AristaFlow BPM Suite { a next generation process management technology that has originated from our research activities in the ADEPT project. We show how AristaFlow ensures ease of use for process implementers, application developers, and end users. Our overall vision is to realize robust and exible process support without encountering bad surprises during runtime. We denote such error-safe process executions as Poka-Yoke Work ows.
For implementing Poka-Yoke Work ows, we pursue the idea of process composition in a \plug & play" style supported by correctness checks. The latter contribute to exclude errors during process execution. As prerequisite, for example, implicit data ow dependencies among application services, which are relevant for their execution order, have to be made known to the process engine. AristaFlow provides an intuitive graphical editor and composition tool to process implementers (cf. Fig. 1), and it applies a correctness by construction principle by providing at any time only those operations to the user which allow to transform a structurally sound process schema into another one; i.e., change operations are enabled or disabled according to which region in the process graph is marked for applying an operation. De ciencies not prohibited by this approach (e.g., concerning data ow) are checked on-the- y and are reported continuously in the problem window of the Process Template Editor.
Generally, we should not require from process implementers that they have detailed knowledge about the internals of the application functions they can assign to process activities. However, this should not be achieved by undermining the correctness by construction principle of Poka-Yoke Work ows. In AristaFlow, all kinds of executables, that may be associated with process activities, are rst registered in the Activity Repository as activity templates. An activity template provides all information to the Process Template Editor; e.g., about mandatory and optional input/output parameters or data dependencies to other activity templates. The process implementer just drags and drops an activity template from the Activity Repository Browser window of the Process Template Editor (cf. Fig. 2) onto the desired location in the process graph. Depending on the inDepending on the intended purpose of usage, such an activity template can be very specific or rather generic. When using a specific activity template, everything may be fixed (the input parameters, the output parameters, all settings, etc.). In this case, the only remaining task for the process implementer is to check whether the proposed mapping of the input and output parameters to process data elements (i. e., the process variables used within this process to communicate among activities) is correct. Using a very specific database actitivity template, in turn, may mean that everyting is fixed: the input and output parameters, the details of the database used and the connection parameters as well as the fully specified SQL statement. A more generic Activity Template may leave open the SQL statement, the number and types of input and output parameters, or the settings for the database connection (in parts or even completely). tended purpose of usage, an activity template can be rather speci c or generic. When us5i.n2g Aachsipeveecmienct: tEeamse pofluasteefoervaepprlyictahtioinngdevcealonpebrse xed; e.g., input/output parametersAasindicated insSeetcttioinn5.1, all application functions atnhdseervicneslyarerreepresaenintedi ningADtEaPsTk2byfoAcrtivtityhe process nd all gs. In this case, o m
Templates. That is any developer who wants to provide a new application function or service will have to impleme nimtpelermeinst atsouitacbhleeacctkivity wtemhpelattheaenrd ptuhtiteintporthoepAcotivsiteydRepmosaitopryp.Tihnisgmaokefs panaarctaivmity teetme-rs to proplate available and accessible within the Process Template Editor during process modeling, as illustrated cess dataineFligeumre 8e.nTotssimipslifyctoherrimepcletm.enWtatihoneonfsuuchsianctigvityatemspplaetecsi, AcDEdPTa2tparobviadesseseavecratlilevvietlsyoftemplate, abstraction. for example, input/output parameters, details of the used database, connection
At the lowest level ADEPT2 provides a so-called Execution Environment for each kind of basic operation parametewrhsic,h AaDnEPdT2tshupeporftsu. lAlmyonsg potehecrsi, AeDdEPTS2 QoffLers esxteacuttieonmenevnirotnmaenrtes for SxQeLdst.ateAmenmts,ore generic web services, EXE files, BeanShell scripts, basic file operations, system-generated forms, etc. However, Activity tTheeimmplepmleanttaeti,onionf antuexrencu,tiomneanvyironlmeeanvtreequiresesnomtehkneowSleQdgeLabsouttaAtDeEmPT2enintternoalrs atnhd,e number op and typettsehemreimfoprele,mweillnttyeprsic.oalluytnopt ubetthpetaasrkaomfaneotredirnsar.y application developer, but will be performed by sysof input/ corresponding activity template and add it to the Activity Repository. It then becomes available in the AristaFlow Process Template Editor for process modeling (cf. Fig. 2). To simplify implementation of activity templates, we support several levels of abstraction. At the lowest one we provide an Execution Environment for each kind of supported basic operation (e.g., SQL statements, web services, EXE les, BeanShell scripts, basic le operations, and system-generated forms). However, implementation of an execution environment requires some knowledge about internals and, therefore, is typically not the task of an ordinary application developer, but is performed by system implementers.
An execution environment de nes the set of methods needed to interact with the runtime system as well as to implement the operations and facilities that shall be provided by the activity template. An activity template for database access, for example, may allow the user to specify connection details. In general, AristaFlow needs information about the runtime behavior of the activities; e.g., whether they may be aborted, suspended, or undone. The developer of an activity template has to implement interface methods that inform the AristaFlow runtime environment which of these facilities are supported. For the latter case he must also provide the implementation of this functionality. The task of implementing a new activity template is simple if it can be based on a generic activity template. In this case, implementation can be reduced to putting the appropriate entries into the set of forms representing the activity template. 2.3
Enabling ease of use for end users is mainly the task of application developers. They decide how \manual" process activities interact with the end user. They also decide whether the standard work ow client is used or whether a dedicated one shall be provided. An important prerequisite for realizing adapted user interfaces is to provide the appropriate methods to the application developer; e.g., to enable him to realize end user interactions in the context of ad-hoc changes. To implement clients with such capabilities, the application developer can make use of system functions available at the AristaFlow API.
Fig. 3 and Fig. 4 illustrate how an ad hoc change may foster Poka-Yoke Work ows. Assume that a process instance wants to issue a request for a book quote using Amazon's web service facilities, but then fails in doing so. The user detects that his process is in trouble and calls the system administrator. The latter then invokes the AristaFlow Process Monitor to take a look at this process instance (cf. Fig. 3). Analyzing the execution log of the failed activity he detects that its execution failed because the connection to Amazon could not be established. Let us assume that he considers this as temporary problem and o ers the user to reset the activity so that it can be repeated once again. Being a friendly guy, he takes a short look at the process instance and its data dependencies, and sees that the result of this and the subsequent activity is only needed when executing the \Choose o er" activity. Therefore, he o ers the user to move these two activities after activity \CheckSpecialO ers"; i.e., the user can continue to work on this process instance before the PAIS tries to reconnect to Amazon. To accomplish this change he would switch to the Instance Change Perspective of the Process Monitor which provides the same set of change operations as the Process Template Editor. In fact, it is the Process Template Editor, but it is aware that a process instance has been loaded and, therefore, all instance-related state information is taken additionally into account when enabling/disabling change operations and applying correctness checks. The system administrator would now mark the nodes \Get Amazon o er" and \Get Amazon price" as source area and the nodes \CheckSpecial O er" and \Choose o er" as target area, and then perform operation Move nodes. The resulting process is depicted in Fig. 4. Another option would be to move node \RetrieveSnailO er" (where we are waiting for an E-Mail response) after \CheckSpecialO er" as well. Then \CheckSpecialO er" would become immediately executable.
Achievement: Complex ad hoc changes and process schema evolution As motivated in in Section 3.4, also complex ad hoc changes must be possible. Figure 11 and Figure 12 illustrate how such a non-trivial ad hoc change could look like. As example, assume that a process instance wants to issue a request for quote for a book using Amazon's web service facilities and but fails in doing so. The user detects that his process instance is in trouble and calls the system administrator for help. The system administrator invokes the ADEPT2 Process Monitor to take a look at this process instance (cf. Figure 11). Looking into the execution log of the failed activity he may detect that its execution failed because the connection to Amazon could not be established. Let us assume, that he considers this as a temporary problem and offers the user to reset this activity so that it can be repeated once again.
Being a friendly guy, he takes a short look a the process instance and its data flow dependencies and sees that the result of this and the subsequent activity is only needed when executing the "Choose offer" activity. He, therefore, offers the user to move these two activities after the activity "CheckSpecialOffers", so that the user can continue to work on this process instance before the PAIS once again tries to connect to Amazon.
therefore, the user may want to call Amazon by phone to get the price that way. In this case we would ask the system adminstrator to delete the two activities in trouble, to replace them with a form-based activity which allows to enter the price manually and which would provide the value for the data element which was previously served by the activity "Get Amazon price".
With respect to process schema evolution, important goals were to allow the full spectrum of change operations, to migrate bFotihg,unroet m11o:diPfierodcaends sinMdivoidnuiatollyr:mModoinfi eitdoprirnocgesPseirnssptaenccetisv(eas far as possible), and to hideF(aitgbe.s3ta.ll) therinohcereesntscoMmploexnityitoofpre:rfoMrmingnailltthoerniencegssaryecrhesckseacstwievlleas the
P o P p In orderretqouidreodsinos,tahnecewsotualtde asdwaiptctihontos athsefaIrnasstapnocsesibClehfaronmgethPeepresrpseonctiinvechoafrgtheetoPpreorcfoersmstMhisontaistokr. Wwheich provides thinevessatemdeaslopteocfterunmergoyfincthoatnhgisearoepaeirnaotirodnesrto find anPcroocmepsreshensmivpelastoeluEtiodnitotor.thIne pfarocbt,leitm'is[R' itnhde04P,rocess
as the Te TemplaRtReDE0d4i,tRoRr.DH0o4wb,eRvReDr,0it4cis], abwutathree trhesautltas phraovceejussstiifniesdt athnecsee heaffosrtbse.en loaded and, therefore, all the instance-related state information is taken additionally into account when enabling or disabling change operations and when performing correctness checks. The system administrator would now mark the two nodes "Get Amazon offer" and "Get Amazon price" as source area and the nodes "CheckSpecial Offer" and "Choose offer" as target area and then perform the operation "Move nodes". The resulting process graph is illustrated in Figure 12. He could also offer to move the node "RetrieveSnailOffer" (where we are wating for an E-Mail response) after "CheckSpecialOffer" as well, then "CheckSpecialOffer" would become immediately selectable (and thus executable). – Assume, the web service problem lasts longer and, ADEPT-CSRD--39.doc
Process Templates
Application Functions
TPermopcleastess AFpupnlcictaiotinosn 3
Process Designer / Process Designer / Due to its \correctness by constrPurocecss Atdmiinistrnator" principle and its abiPlroictess yAdministorator quickly como t a) Process schema chanagend adaptable pb)rCohceceksstsaeteso,frunning instances pose robustly executable AristaFlow is suited to enable Poka-Yoke (= error-safe) Work ows. Currently, we apply AristaFlow in di erent industrial context with very positive feedback on its components and features.