Pull Deployment of Services

The NWO/EZ Jacquard Software Engineering Program has granted the project Pull Deployment of Services for an amount of 368K Euro which should pay for a PhD student (4 years) and a postdoc (3 years). In the project for which I am principal investigator, we collaborate with Merijn de Jonge from Philips Research and the buildfarm project at TU Delft in which software deployment expert Eelco Dolstra is postdoc. Here's the text from the proposal summary:

Hospitals are complex organizations, requiring the coordination of specialists and support staff operating complex medical equipment, involving large data sets, to take care of the health of large numbers of patients. The information technology infrastructure of hospitals is heterogeneous and may consist of thousands of electronic devices, ranging from workstations to medical equipment such as MRI scanners. These devices are connected by wired and wireless networks with complex topologies with different security and privacy policies applicable to different nodes. Software deployment in such a heterogeneous environment is inherently difficult. In order to make health-care professionals more effective and deployment and maintenance more tractable, the hospital information technology infrastructure is changing from a device-oriented to a service-oriented environment, in which the access to services is decoupled from the physical access to particular devices.

In this project, we propose a pull model for service deployment in which the components comprising a service are distributed over nodes in the network, depending on the network topology, properties of the application, and quality of service requirements. The goal of this project is to expand the state-of-the-art in software deployment to support pull deployment of services. In order to realize this goal we will conduct research in (1) modeling of services and network architectures, (2) technology for distributed deployment, and (3) tools for testing implementations of distributed services. We will build on our previous research in software deployment (Nix) and model-based software development (Stratego/XT). The project will be conducted in close collaboration with Philips as industrial partner and will consist of a series of experiments building prototype systems which implement service distribution scenarios of increasing complexity.