RAINDROP: Cross-stack Adaptation for the Edgification of Microservices Applications

RAINDROP seek to address in a unique way the challenge of building, deploying, and operating applications on upcoming cloud infrastructures formed of a continuum of core (data centers) and edge resources. The project targets the demanding class of collaborative and latency-sensitive applications for which the use of edge resources close to users should allow interactions and collaborations with greatly reduced latencies and increased responsiveness, enabling new classes of usages such as collaborative work in near-real-time, virtual spaces, or smart environments. The project focuses on application back ends built using service-oriented architectures and more specifically the micro-service paradigm.

RAINDROP started in September 2022. The project is jointly funded by UCLouvain and UNamur as part of the ARC (Actions de Recherche Concertées). It involves the Cloud and Large Scale Computing group, the Security and Performance of Networked Systems at UCLouvain (PI: Pr. Ramin Sadre) and the Research Center on Information Systems Engineering (PReCISE) at UNamur (PI: Pr. Anthony Clève).

Objectives

RAINDROP leverages a unique collaboration between software and data engineering, cloud platforms and runtimes, and infrastructure management. It will allow application designers to build and gradually adapt microservices applications from a cloud (data center) target towards novel distributed cloud infrastructures such as the edge-cloud continuum and multi-site (micro-cloud) environments.

At the heart of the project will lie the synergy between three complementary forms of adaptation:

  • for applications deployment, placement, and configuration during their execution,
  • for the support infrastructure under dynamic demand and availability of resources,
  • and through the necessary software evolution to make applications increasingly able to benefit from such automated runtime deployments over the edge-core continuum.

The interoperation between these different classes of adaptations will leverage common application and infrastructure models, and enable a feedback loop between runtime observations and developers enacting software evolution.

RAINDROP will leverage best practices in cloud application development and operation, and intend to significantly improve programmability and abstraction for cloud developers wishing to take advantage of the promising, but demanding and complex novel core-edge infrastructures.

The overall vision of RAINDROP is illustrated by the figure below.

RAINDROP overview

Participants

The following persons are involved in the project:

  • [UCLouvain] Prof. Etienne Rivière, spokesperson for the project and leading the activities around application adaptation;
  • [UCLouvain] Prof. Ramin Sadre, leading the activities around infrastructure adaptation;
  • [UNamur] Prof. Antony Clève, leading the activities around software evolution and data management adaptation;
  • [UCLouvain] Carole Djeumo Noucti is a Ph.D. student, focusing on runtime placement and adaptation of microservices applications across the edge-cloud continuum;
  • [UCLouvain] Yinan Cao is a Ph.D. student, focusing on infrastructure management and adaptation in the continuum;
  • [UNamur] Maxime André is a Ph.D. student working on software evolution and data management for microservices applications.
  • [UNamur] Dr. Loup Meurice is a post-doctoral researcher working on software evolution aspects.
  • [UCLouvain and UNamur] In 2023, the project will offer a post-doctoral fellowship to work on cross-stack adaptation and on the joint modeling of infrastructure and applications in the edge-cloud continuum. If you currently are a Ph.D. researcher outside of Belgium with potentially matching research interest, contact Prof. Etienne Rivière.

In addition to its core members, RAINDROP intends to establish collaborations at the Belgian and international levels. External members are warmly encouraged to get in touch with either of the co-PIs for any question or collaboration request.