MSc Defence by Kasper Myrtue Borggren

Title

Scalable Structural Health Monitoring Data Platform using Actors as a Database

Abstract

AMEG is a company that creates Structural Health Monitoring Systems for bridges, factories, and other engineered structures. Currently the company builds a custom data platform for each customer, with interaction between sensors, and users. We investigate how a multi-tenancy system can be created that can ingest data from all their clients, with the same requirements as earlier AMEG projects for functionality and user interaction. We find that using Orleans virtual actors as a database, together with a data warehouse, should be able to address all of AMEG's requirements. Using Orleans virtual actors as a database raises certain issues with modelling the database, navigating the database without indexing, and the lack of ACID-compliant transactions across virtual actors. In order to design a database structure for Orleans virtual actors, we extend the ER diagram model to accommodate for actors. We implement a solution in the Orleans virtual actor framework, leaving the data warehouse to be integrated another time. In order to prove that the solution can handle an increasing amount of clients, we benchmark the Orleans virtual actor implementation to show that it can scale out. In addition, we show that the implementation can handle user interaction, and sensors sampling data at different frequencies. We find that using Orleans virtual actors as a tenancy system for AMEGs clients will be able to scale out, handle a variate of sensor data sampling frequencies, and user interactions.

Supervisors:
Marcos Antonio Vaz Salles, DIKU
Rasmus Brøndum, AMEG

External examinator:
Czeslaw Kazimierczak, CSC Danmark A/S