Section: New Results

Data Centered Service Oriented Computing

Business process distribution on a SaaS architecture

Participants : Walid Fdhila, Claude Godart, Elio Goettelmann, Samir Youcef.

The objective of this work is to support the deployment of a business process as a set of distributed services provided partially or totally off-premises or even in the cloud. Direct applications in our target are:

  • A methodological approach for choreographies elicitations and monitoring[12] .

  • An algorithm for optimized service providers selection (including cloud) [11] , [9] , [10] .

In this objective, we have deployed two approaches. A first is based on heuristics (greedy algorithm to compute an initial solution, combined with a tabu search) for optimizing the selection of services assigned to activities in a decentralized composite service, both in terms of the overall QoS of the composite service and the communication overhead; in output, the initial business process model is translated in a set of interconnected business process fragments.

A second approach uses operational research techniques for optimizing a cloud selection taking into account two conflicting objectives, namely: the execution time (makespan) and the overall cost incurred using a set of resources. We proposed in [9] three complementary approaches to deal with the matching and scheduling scientific workflow tasks in Cloud computing environments. An extension of this first study was presented in [11] , [10] . More precisely, we have extended the three proposed approaches to consider: (i) the business workflows and (ii) the concurrent access to resources by multiple instances of a given process. To achieve this goal, we proposed to use a predictive models in order to estimate the availability of the used resources. We are currently working on the business processes execution in Cloud computing context taking into account workflow patterns such as sequence, switch, multi-choice, etc. patterns. Moreover, we plan to extend the proposed work to take into account others criteria like carbon emission and energy cost.

Alignment between Business Process and Service Architecture

Participants : François Charoy, Karim Dahman, Claude Godart.

In the continuation of work done previously on change management during process execution, we are conducting work on the governance of change at the business level and on its implications at the architecture and infrastructure level of an information system. Last year was devoted to the definition of the transformation rules that allowed to go from a business model to an IT model, i.e. a transformation between model based on different paradigms. During this year, a great deal of effort has been done in order to extend our work on Business to IT alignment management. Our goal is still to maintain this alignment at the lowest possible cost when the business process are changing. Further than that we are trying to describe and validated an engineering method to help designer to maintain this alignment. Karim Dahman has defended is PhD on this matter in october 2012.

Monitoring and violations detections of choreographies or distributed compositions of services

Participants : Aymen Baouab, Ehtesham Zahoor, Olivier Perrin, Walid Fdhila, Claude Godart.

The dynamic nature of the cross-organizational business processes poses various challenges to their successful execution. Services choreographies or distributed compositions of services help to reduce such complexity by providing means for describing complex systems at a higher level. However, this does not necessarily guarantee that erroneous situations cannot occur due to inappropriately specified interactions. In [7] , [6] , we propose an approach for decentralized monitoring of cross-organizational choreographies, using a runtime event-based approach to deal with the problem of monitoring conformance of interaction sequences. Our approach allows for an automatic and optimized generation of rules. After parsing the choreography graph into a hierarchy of canonical blocks, tagging each event by its block ascendancy, an optimized set of monitoring queries is generated. We evaluate the concepts based on a scenario showing how much the number of queries can be significantly reduced. These results use our previous results about event-based framework DISC [33] .