Towards Sustainable Solutions for Applications in Cloud Computing and Big Data

Date(s) - 05/04/2017
2:00 pm - 3:00 pm


Nguyen Ho Thi Thao

Abstract: The rapid growth of energy consumption and consequently the CO2 emission of ICT sector have raised serious concerns about their environmental impact. In data centers, recent research in energy efficiency has been focusing on reducing the energy consumption, and state-of-the-art techniques have been emphasized on optimizing power and energy consumption at hardware and infrastructure levels. This talk provides another perspective to further improve energy efficiency at the level of applications in the context of Cloud Computing and Big Data. Particularly, it proposes to tackle the energy efficiency problem from two different angles: the possibility to reduce the energy consumption of applications and the potential to reduce the amount of data to be processed.

Adopting both modeling-based and experimental-based techniques, the first approach proposes analytical models to estimate the energy consumption of cloud-based applications and quantify the energy consumed per job. The models are built based on cloud infrastructure’s and application’s parameters and are validated experimentally using a real cloud infrastructure. The energy per job metric is used as the main driver to analyze energy consuming behaviors of applications and to assess possibilities to improve their energy efficiency. Through the selected case studies, we demonstrate that application’s energy efficiency can be further improved at the application level through their execution configurations and/or through optimizing load distribution. As a final result, the proposed method provides a means to both application’s users and cloud infrastructure’s providers to actively control and optimize the energy usage of cloud-based applications.

In the second proposal, we propose a novel approach for developing sustainable applications in the context of Big Data by considering the potential to reduce data volume. In particular, data volume reduction can be achieved through identifying and removing irrelevant data with respect to user’s goals. For doing this, we propose a correlation-based method to analyze the relationship between different data sets and based on that to understand if considered data are relevant to given goals. We develop a scalable and efficient algorithm to analyze the relationships among spatial-temporal big data sets, and test the performance of our method on real world data sets.

Short Bio:
Nguyen Ho Thi Thao received the PhD in Computer Science at Politecnico di Milano, Italy in Feb 2017. Previously, she was a visiting research scholar at New York University, Center of Urban Science and Progress. She obtained the master degree in Computer Science with distinction (cum laude) from Politecnico di Milano and the bachelor degree in Computer Science from Ho Chi Minh University of Technology, Vietnam.

Her current research focuses on energy efficiency solutions for cloud infrastructures and applications. Recently, her research shifts to the emerging Big Data systems and applications, with initial interest in analyzing spatial-temporal big data sets. Her research interests cover a wide range of disciplines, including machine learning, big data analytics and data science, database and system optimization.