Dynamic Web Services Selection in Service-Oriented Distributed Environments

Abstract

Nowadays, the Service-Oriented Distributed Systems are de facto platforms for service providers exposing functionalities to external clients. This type of environments allows distributed deployments of web services with the same functionality to multiple locations (called service replicas) in order to fit the clients' requirements. On consumer's side, accessing these services, in a context that gathers quality and business constraints, is most of the time difficult regarding identification of the most suitable service replica. The current work focuses on defining a model and a recommender system logical architecture that help the clients to find the most effective service replica for their needs. Also, the solution serves service providers in their goal of improving the quality of services for their clients. The model combines quality criteria (e.g. processing time), business constraints (e.g. execution price), and reputation while taking into account the caller's location, replica's location, and the ability of the computing systems hosting replica to serve a request under a given load. We target our model to those organizations having critical Straight-through Processing, Extreme Transaction Processing applications with associated business constraints and where the quality of the external services impact directly clients' SLAs.