Résumé: Data synchronization and content delivery services are key to supporting healthcare dataflows built by organizations. These types of services must prepare and process the data to accomplish mandatory non-functional requirements, such as security and reliability. This is a challenge as multiple applications, infrastructures, and platforms participate in healthcare dataflows. This paper presents FedFlow, a federated content distribution platform to build infrastructure-agnostic health data sharing and synchronization services to support healthcare dataflows. FedFlow creates secure and efficient data sharing and synchronization patterns for intra-dataflows and inter-dataflows by using implicit parallel data preparation schemes. A prototype of FedFlow was developed to conduct a case study about the building of inter-dataflows for delivering synchronized health data to multiple organizations by using combinations of non-functional requirements algorithms to accomplish governmental rules related to health data management. The experimental evaluation in a multi-cloud federated environment showed that FedFlow is around 90% faster than a traditional pipeline implementation, around 40% faster than Jenkins workflow management, and almost 30% faster than duplicity.

Langue: Anglais