Composing data services in a dynamic way
We investigate models, algorithms and tools for coordinating services with non functional properties (contracts) and for providing access to heterogeneous data coming from services.
List of participants: Ch. Bobineau (Associate PR), Ch. Collet (PR), G. Vargas-Solar (CR1), Javier Alfonso Espinosa-Oviedo (PhD 2009 - , co-direction France - Mexico), Alberto Portilla-Flores (PhD 2005- 2009 co-direction, France - Mexico), Victor Cuesvas-Vicentt´ın (PhD 2006 - )
Composing services exported by different organisations is a key issue when building large scale and data-intensive applications/systems. It is becoming more crucial when considering services within ubiquitous infrastructures made of heterogeneous devices, servers, applications connected with heterogeneous systems. Composition requires to take into account the characteristics of these eco-systems (e.g., memory and computing, and network capabilities). The composition process uses this knowledge or semantics to dynamically discover and coordinate (ubiquitous) services, and then to adapt the coordination process depending on the availability and change of services. Another important challenge is to consider non functional aspects and QoS (quality of service) criteria such as availability, reliability, and temporal constraints that are crucial when composing data services in a dynamic way. Int he group, we investigate models, algorithms and tools for:
- merging data and control flows for describing service coordination ;
- processing (accessing and integrating) heterogeneous data coming from services in a discrete and continuous way;
- reliable and adaptive data services composition;
- dealing with autonomic services and systems.
We plan to extend our work on reliable and autonomic data services composition in three directions:
- Providing reliability to services composition: we will define a language for specifying non functional properties of services’ coordinations. In particular, it will be used for programming recovery actions with associated execution strategies for synchronizing recovery with the execution of the coordination. The semantics of the language will be formally described and its properties will be studied and demonstrated (ECOS-ANUIES ORCHESTRA project).
- Service-based query processing: we will investigate new query processing techniques that tackle at the same time classic, mobile and continuous queries by composing data services providers that are (push/pull, static or nomad) providers (OPTIMACS ANR project, PhD). At the opposite we will investigate how to use declarative queries expressions in integrating and composing services available in dynamic environments (e-CLOUDSS and redSHINE projects).
- Event flow management: research on events and rules will continue as a base for the autonomic management of views for data spaces and services clouds. It will address event management through a flexible approach that enables programmers to build their own event composition/synthetisis and management functions.