GMAP Member Presents his Research Proposal for his Ph.D. Dissertation (Adriano Vogel)

On Friday, November 22, 2019, Adriano José Vogel, had his thesis proposal approved. Adriano is a doctoral student in the Graduate Program in Computer Science at the Polytechnic School of the Pontifical Catholic University of Rio Grande do Sul (PUCRS). The work was supervised by Dr. Luiz Gustavo Fernandes and co-supervised by Dr. Dalvan Griebler. The members of the examining board were Dr. Fernando Luís Dotti (PPGCC/PUCRS) and Dr. Rodrigo da Rosa Righi (PPGCA/UNISINOS).

Following is the title and summary of the thesis proposal:

Title: Towards Efficient and High-level Parallelism Abstractions for Stream Processing with Self-adaptivity

Abstract: Stream processing is a representative paradigm present in several applications that compute data flowing in the form of streams (e.g., video feeds, image, and data analytics). However, processing data and producing results periodically is a technological challenge. As a consequence, the majority of these applications demand parallelism for accelerating their executions. Although high-level and structured parallel programming aims at facilitating parallelism exploitation, there are still issues to be addressed for improving existing parallel programming abstractions. The complexity increases when application developers must set non-intuitive and error-prone parallelism parameters. Employing self-adaptivity in stream processing applications can provide a higher level of programming abstractions and autonomic resource management. However, providing highly abstracted solutions that are efficient and performatic can be challenging. Relevant and novel opportunities for selfadaptive parallelism are concerned to support adjustments in complex application compositions, which are representative for real-world applications. There is also a demand for new methodologies and benchmarking metrics for comprehensively measuring the impact of self-adaptivity in application performance and resource usage. Moreover, a further opportunity for increasing flexibility and performance is dynamically changing the application graph’s topology. The preliminary results have shown that self-adaptation can achieve additional parallelism abstractions for running stream processing applications. Also, the preliminary results from proposed strategies have shown a competitive performance and a low overhead.