Programme INRIA "Equipes Associées"
/ INRIA "Associate Teams" Programme

EQUIPE ASSOCIEE / ASSOCIATE TEAM	FOSSA: Formalizing Orchestration & Secure Services Analysis
sélection	2010

 

Equipe-Projet INRIA : DISTRIBCOM	Organisme étranger partenaire : University of Texas at Austin
Centre de recherche INRIA : Rennes - Bretagne Atlantique Thème INRIA : Réseaux, systèmes et services, calcul distribué	Pays / Country : United States of America

 

	Coordinateur français / French Coordinator	Coordinateur étranger / Partner Coordinator
Nom, prénom / First name, Given name	Benveniste Albert	Misra Jayadev
Grade, statut / Position	Directeur de Recherche INRIA	Professor and Schlumberger Centennial Chair in Computer Sciences
Organisme d'appartenance / Home Institution	INRIA/IRISA	University of Texas/Dept. of Computer Sciences
Adresse postale / Postal address	Campus de Beaulieu, 35042 Rennes, France	Dept. of Computer Sciences, Taylor Hall, Room 3.102, Univ. of Texas at Austin, Austin, Texas 78712-1188, USA
URL / Website	http://www.irisa.fr/distribcom/benveniste	http://www.cs.utexas.edu/users/misra
Téléphone / Telephone	+33 2 99 84 72 35	+1 512-471-9550
Télécopie / Fax	+33 2 99 84 71 71	+1 512-471-8885
Courriel / Email	Albert.Benveniste@irisa.fr	misra@cs.utexas.edu

Equipe-Projet INRIA : MEXICO
Centre de recherche INRIA : Saclay - Ile de France Thème INRIA : Programmation, vérification et preuves

 

	Autre participant français / Other french Participant
Nom, prénom / First name, Given name	Haar Stefan
Grade, statut / Position	Chargé de Recherche INRIA
Organisme d'appartenance / Home Institution	INRIA/LSV
Adresse postale / Postal address	ENS de Cachan, 61, avenue du Président Wilson, 94235 CACHAN Cedex, France
URL / Website	http://www.lsv.ens-cachan.fr/~haar/
Téléphone / Telephone	+33 1 47 40 75 67
Télécopie / Fax	+33 1 47 40 75 21
Courriel / Email	haar@lsv.ens-cachan.fr

Analyse formelle de l'orchestration de services / Formalizing Orchestration & Secure Services Analysis (FOSSA*)

Descriptif / Description :  The widespread deployment of networked applications and adoption of the internet has fostered an environment in which many distributed services are available. There is great demand to automate business processes and workflows among organizations and individuals. Solutions to such problems require orchestration or choreography of concurrent and distributed services in the face of arbitrary delays and failures of components and communication. The Orc team, lead by Jayadev Misra at the University of Texas at Austin, has developed the Orc language to support orchestrations. The DistribCom team has developed studies regarding the Quality of Services of orchestrations and choreographies, with emphasis on Orc. Finally, from the newly created (2009) MExICo team in Saclay ( http://www.inria.fr/recherche/equipes/mexico.en.html), Stefan Haar is a former member of DistribCom and has participated in the above research, and Serge Haddad has been working on client synthesis and aspects of orchestration, in particular adaptation. The teams cooperate since 2006 and have decided to join their efforts in lauching the associated team FOSSA, with the following objectives: To contribute to the development of Orc [2] as a support for Structured Application Development over Wide-Area Networks; To develop a comprehensive theory of QoS for composite Web services, supporting: SLA contracts, contract composition, contract monitoring, and reconfiguration; To experiment on real orchestrations or choreographies; To develop studies on security (Authorization and Information flow); To develop studies on the functional synthesis and design of composite services, including mashups ("à la Hop" [5]) and synthesis of adaptors for services inside a composition; To develop the synthesis of clients supporting the interaction protocol of a composite service; To benchmark different styles of formalisms: Orc, a graphical formalism for workflow specifications by Gero Decker (Signavio), and Active XML document based formalism [1]. As a general umbrella for all the above objectives, distributed aspects will be central. (*) FOSSA is an endangered animal (http://lefossa.org.free.fr/Animal/description.htm). It is ferocious and enjoys a long life.

Présentation détaillée de l'Équipe Associée
Detailed presentation of the Associate Team

1. Objectifs scientifiques de la proposition
/ Scientific goals of the proposal

Background of the collaboration between the teams: In 2002, Albert Benveniste attended as a plenary speaker the Europar Confer- ence. Jayadev Misra gave another plenary talk where he presented his early view on wide area computing that subsequently lead to the birth of Orc. In 2005, when Sidney Rosario joined the DistribCom team for his Master degree, contact was established with the group of J. Misra regarding Orc. The first cooperation between the two groups was on semantic issues regarding the Orc language [8]. Since then, regular contacts have been maintained with visits from both sides. The two teams decided in 2008 to join their efforts at developing QoS studies for Orc.

As a result of this past and present interaction, the thesis of Sidney Rosario (defense to be held on November 30th) nicely combines the following two aspects:

• Studies on QoS for orchestrations, covering QoS contract composition and monitoring using probabilistic and statistical techniques; the results of this thesis are both fundamental in nature and effective through a prototype software tool.

• Development of this approach on top of the Orc language and execution engine as provided by UT Austin.

William Cook is a committee member of the PhD thesis of Sidney Rosario and Sidney will be a post-doc at the Orc group starting end of october (sub ject to visa issues). During his post-doc at Austin, Sidney will continue and complete the development of a QoS management service on top of Orc. Deeper interaction with the Orc group at Austin will, however, be instru- mental in achieving the above mentioned wider program. This motivates the application for an Associated Team grant.

The teams:

• DistribCom team at INRIA: DistribCom is a joint team of INRIA, CNRS, Université of Rennes I, and ENS Cachan, located at IRISA, Rennes. Today, research on network and service management as well as Web Services mainly focuses on issues of software architecture and infrastructure deployment. However, these areas also involve algorithmic problems such as fault diagnosis and alarm correlation, testing, QoS evaluation, negotiation, and monitoring. Our group focuses on these issues. More precisely, we address models and algorithms for distributed network and service manage- ment, and the distributed management of Web services and business processes. Our main industrial ties are with Alcatel-Lucent, on the topic of networks and service management. The 2008 activity report of the team is available from http://www.inria.fr/recherche/equipes/distribcom.en.html. The researchers involved in the associated team are the following:

  • Albert Benveniste will be the coordinator of FOSSA. He is Directeur de Recherche at INRIA. His current interests include: system identification for vibrations mechanics, component and contract based design of em- bedded systems, real-time architectures, network management, Quality of Service management of Web services, and distributed active XML docu- ments. In 1980 Albert Benveniste was co-winner of the IEEE Trans. on Automatic Control Best Transaction Paper Award for his paper on blind deconvolution in data communications. In 1990 he received the CNRS silver medal and in 1991 he has been elected IEEE fellow. In 2008 he was winner of the Grand Prix France Telecom of the french Academy of Sciences. From 1986 to 1990 he was vice-chairman of the IFAC commit- tee on Theory and was chairman of this committee for 1991-1993. He has been or is member of the Editorial Board of several journals, and in particular IEEE Transactions on Automatic Control and the Proceedings of the IEEE. Since 1997, he has been responsible for INRIA of the joint Alcatel-INRIA research programme and is now chief scientist of the joint Bell Labs-INRIA research lab. He is member of the scientific advisory board of SAFRAN Group.

  • Claude Jard graduated from Ecole Nationale des Telecommunications, Paris, in 1981. He obtained his Ph.D. in Computer Science from the university of Rennes in 1984. He was the head of the protocol valida- tion group at the French National Research Centre in Telecommunications (CNET) from1981 to 1986. He has been a Research supervisor, and then a Research director at the French National Centre for Scientific Research (CNRS) since 1987. Currently, he is professor in Computer Science at the Ecole Normale Supérieure de Cachan. He is the Director of the research of the Brittany branch of this school. Prof. Jard's research works relate to the formal analysis of asynchronous parallel systems. His current interests are the non-functional aspects like time and QoS in large scale systems. He is the author or co-author of more than 140 publications, carried out pri- marily within three research communities: theoretical computer science, protocol engineering, and distributed systems.

  The following post-docs and students will participate in FOSSA:

  • Benoît Masson, post-doc INRIA on the CREATE-ACTIVDOC for one year starting from October 15, 2009. Benoît Masson will mainly work with Loic Hélouët and Albert Benveniste on Active XML. In the context of FOSSA, he will participate to the evaluation of AXML for the selected applications.

  • Ajay Kattepur, bourse INRIA-CORDIS (2009-2012). Ajay Kattepur graduated from Nanyang Technological University (NTU) Singapore Master of Engineering (Research) student in the School of Electrical and Electronics Engineering, Division of Information Engineering. His PhD subject takes the thesis of Sidney Rosario [3] as a background and will be focused on the objectives of FOSSA.

  • A joint post-doc in the framework of FOSSA.

• The Orc group, University of Texas at Austin: The Orc group has the research about and development of the Orc language in its focus. The researchers from this group involved in the associated team will be:

  • William Cook, Professor at University of Texas at Austin. William Cook is an Assistant Professor in the Department of Computer Sciences at the University of Texas at Austin. His research is focused on object-oriented programming, programming languages, modeling languages, and the in- terface between programming languages and databases. Prior to joining UT in 2003, Dr. Cook was Chief Technology Officer and co-founder of Allegis Corporation. He was chief architect for several award-winning products, including the eBusiness Suite at Allegis, the Writer's Solution for Prentice Hall, and the AppleScript language at Apple Computer. At HP Labs his research focused on the foundations of object-oriented lan- guages, including formal models of mixins, inheritance, and typed models of ob ject-oriented languages. He completed his Ph.D. in Computer Science at Brown University in 1989.

  • Jayadev Misra, Professor and Schlumberger Centennial Chair in Computer Sciences at University of Texas at Austin. Misra's research interests are in the area of concurrent programming, with emphasis on rigorous methods to improve the programming process. He is currently spear- heading an effort, jointly with Tony Hoare, to establish a grand challenge project to automate large-scale program verification. Misra is a fellow of ACM and IEEE; he held the Guggenheim fellowship during 1988-1989. He was the Strachey lecturer at Oxford University in 1996, and he held the Belgian FNRS International Chair of Computer Science in 1990.

  • Sidney Rosario, Post-Doc (Post-Doc, supported by a NSF Grant NSF CCF-6752487). Sidney Rosario was a former PhD student at DistribCom team. He performed his thesis on the first two topics of the objectives of the project.

  The following students will participate to the FOSSA:

  • David Kitchin

  • John Thywissen

  • Amin Shali

  • (and possibly one more)

  We project that one or more of the UT Austin students will be jointly supervised by other members of the associated team. This will build on our existing collaboration. This will be formalized by inviting members of the team to join the students thesis committees. None of the UT Austin students listed above have yet formed their committees, so the timing of the collaboration is good.

• The Mexico team at INRIA : The MExICo team at INRIA Saclay/ENS de Cachan has been recently cre- ated under the direction of Stefan Haar, who transferred from the DistribCom team where he had already participated in the cooperation. MExICo's focus is the Modelling and Exploitation of Interaction and Concurrency. The team's interest include concurrent and compositional web service orchestrations in general, and more specifically the tasks of Adaptation and Grey Box Management: In the context of web service orchestrations, existing service components need often to be adapted to new offers or demands, to be replaced in case of disfunction, etc. This integration of components into an existing orchestration of web services requires to adapt services to changing network conditions or user preferences. The task of adaptation thus comprises the detection of violations for specifications or contracts, the identification of the component to be replaced and the new service to replace it, and most importantly steering the interaction with the components already in place to ensure compliance with the intended compound behaviour. Since the components are non-proprietary and their implementations in general only partly known, adaptation thus becomes the task of manageing the interaction behaviour of boxes that are known only through their public interfaces; for their semi-transparent-semi-opaque nature, we denote them as grey boxes.

• Serge Haddad is a former student of ENS Cachan in mathematics. After the doctorate from University of Paris 6 in 1987, he became associate pro- fessor at Paris 6 in 1988. Then he obtained a position of full professor at University Paris-Dauphine in 1993 and finally joined ENS Cachan in 2008. Serge Haddad's research interests concern the design, the verification and the evaluation of distributed systems and cooperative applications includ- ing timed requirements. His favourite formal models are Petri nets (he's member of the steering committee of the conference ATPN), process al- gebra, automata products and their timed, probabilistic and stochastic versions. Recently he focused on quantitative verification, a promising research direction, and obtained significant results ("best paper award" 2005 and 2007 at the conference QEST mainly devoted to this topic).

• Stefan Haar studied mathematics at Hamburg University, Germany, and Johns Hopkins University, Baltimore, MD. He received the Diplom (M. Sc.) in mathematics - where he specialized in stochastic processes - and PhD degrees in computer science (Petri net theory) at Hamburg Univer- sity. After post-doctoral positions in Berlin, Nancy and Paris, he has been a researcher with INRIA Rennes from 2001. In 2007, he went for a sabbatical stay to University of Ottawa, Canada, and was with ALU Bell Labs Ottawa for the first half of 2008; on returning to France he moved to INRIA Saclay at Ecole Normale de Cachan. Since 2001, he has been consistently involved in telecommunication related research, in particular fault diagnosis, quality of service contracts, and web service orchestration; he has been an associate editor of IEEE Transactions of Automatic Con- trol since 2005. His research interests include Petri nets, event structures and graph grammars, conformance testing and probabilistic discrete event systems.

3. Impact
/ Impact

On the scientific objectives of participating teams: The DistribCom team wishes to revisit its research ob jectives regarding Web services orchestrations and QoS studies. Its past research on this topic suggests that the wide scientific background of its members (including statistics and control science, which is part of the new PhD student Ajay Kattepur) should be extremely helpful in addressing the cross-disciplinary issues related to Web services. Performing best investment based on this background is a key issue for the group. This requires a deeper understanding of practical problems when developing applications in the targeted sectors. And this is only possible if the research can build on a tool environment where real-size experiments can be performed and research results can be fielded. The MExICo Team intends to intensify research on grey-box management in the context of orchestrated web services. One tier of its participation is continued implication in the QoS-related research. Indeed with its solid skill in performance evaluation, the team wants to develop dedicated techniques for efficient measurements and improvement of QoS for Web services. Besides this, MExICo aims at developping, instanciating and integrating methods for adap- tation of services into existing compositions. Previous work in this direction had been oriented towards the Software-bus environment; integration with the Orc engine is a desideratum for MExICo. Finally, MExICo strives to improve and facilitate aspect-oriented programming in Orc' wide-area setting as it helps to manage the complexity of the business logic. Regarding the Orc group at UT Austin, the goal of the Orc project is to develop a robust model of structured concurrent programming. The Orc Project is funded by two major grants, one from NSF and another large government grant.

On the relationships between partners and their home institutions:

• Distribcom at INRIA: Research of the DistribCom team on the topics of FOSSA are within the core of the research agenda of the team as submitted when launching it and also at the last team evaluation in 2008. Specific supports for this topic include the following past or present grants or contracts:

• RNRT-SWAN large cooperative pro ject was held in 2005-2008, with several partners including Alcatel and France-Telecom. During this project the research on QoS and orchestrations was launched and sustained since then.

• INRIA Action de Recherche Coordonn´ee ASAX (2005-2007), regarding the distributed systems aspects of Active XML documents infrastructure proposed by Serge Abiteboul and his team at INRIA. Partners of ASAX were the team of Abiteboul, the team of Anca Muscholl in Liafa, Paris VI, and DistribCom.

• ANR-DOCFLOW grant (2006-2009), with the same set of participants and a similar objective. The participation of DistribCom team in this research is also supported by a significant CREATE-ACTIVDOC grant of the Région Bretagne.

Still, this research area of DistribCom lacks industrial interaction for its strategic choices. The important telecommunications industry in France is not the best partner. Better partners are found in IT industry sector, where major actors are IBM and SAP. We hope that the research developed by FOSSA will provide us opportunities for better industrial interactions in this sector.

• Mexico at INRIA: MExICo shares DistribCom's history and research interests, and is furthermore interested in service adaptation and client synthe- sis. Specific supports for the FOSSA topics include the following past or present contracts:

• Research agreement about formal methods for software engineering be- tween the team of S. Haddad (formerly in Univ. Paris-Dauphine) and the research agency FAPEMA of the State of Maranhao in Brazil (2006-2007). The main topics was adaptation in Web services and has led to several joint publications.

• ANR-DOTS (2007-2010) with LABRI, IRCCYN, IRISA and LAMSADE as partners. Complex systems, such as embedded systems that are widely used nowadays (telecommunication, transport, automation), are often dis- tributed (composed of several components that communicate together), timed (contain timing constraints), and open (interact with their environ- ment). Each of these aspects considered separately is now relatively well understood and corresponds to an active research area. The big challenge is to deal with systems which present several of these feature. In the con- text of FOSSA, this corresponds to take into account timing requirements in the interface of services during composition.

• ORC group at UT Austin: The goal of the Orc project is to develop a robust model of structured concurrent programming. The Orc approach supports orchestration of concurrent and distributed services in the face of arbitrary delays and failures of components and communication. This overall goal is divided into three parts: handling communication and time in distributed systems, enhancing the model to allow yet more abstractions for structuring, and implementing Orc with integration to existing languages. The results from these research activities will be evaluated for theoretical completeness and practical usability. The project builds on the existing body of research on concurrent systems while leveraging the PIs' extensive experience with distributed systems, verification, semantics, and language design. The broader impact of this research is to simplify concurrent programming.

Description du programme scientifique de travail (1 à 2 pages maximum)
/Description of the scientific work programme (maximum 1 to 2 pages)

The work program is structured according to the successive years and contribution by the partners. The steps that we detail below will constitute the concerted effort of FOSSA. One should, however, remain aware that the work programme below will be complemented by more opportunistic research around the same topics, depending on the new and unexpected issues and advances that will emerge during the project's period. Also, since Orc is the tool that is put in forefront of this project, contributions to and enhancements of Orc will be performed in an opportunistic way all along the project period. Those invariant aspects of the work programme will not be further mentioned below.

The first year will be mainly devoted to the exploration of realistic size orchestration or choreography applications. Several formalisms will be experimented, including (of course) Orc. Target sectors will be business processes, logistics applications, and, with lesser priority, consumer applications. Aim will be to gather deeper understanding fo the variety of issues underpinning the development of such applications. The ob jective here is to avoid addressing fake issues or issues of minor importance for the targeted sectors. The three partners of FOSSA already have some experience of this kind. In particular, in their past cooperation, DistribCom and UT Austin have studied medium size orchestrations as encountered, e.g., in travel planning. In a different context, DistribCom has studied the whole system supporting the sale and manufacturing of Dell computers, ranging from the Web based front-end for gathering customer orders, to the supply chain involving the distributed interaction of Dell with its suppliers. Emphazis for this first year will thus be on objectives 3 and 6. As a result, key research challenges will be revisited and made more precise. One important additional point is the enhancement of Orc with QoS features. This will be performed as follows. As part of the background of FOSSA we have developed an instrumented semantics of Orc enabling the on-the-fly evaluation of QoS parameters, for an Orc program. This will be further studied and developed. The above approach leaves open the option as to whether a loosely or tightly coupled instrumentation of Orc will be designed.

Central effort of the second year will be the research aspects as prioritized during year 1 of FOSSA. Consequently, focus for this year will shift to ob jectives 2, 4, and 5. During this year, the decision of which formalisms for orchestrations or choreographies FOSSA will continue investigating, will be finalized.

The third year will focus on the tool development resulting from the advances in research performed during year 2 of FOSSA. Research results will be turned into tools. Packaging of the resulting tools and services will be performed as Orc services. Roles of the partners will be made precise after Year 1.