Big Data analytics middleware

The High-Performance Systems group is working on the development of next-generation middleware solutions for Big Data analytics. We aim at exploring new models for distributed and large-scale data analysis and at designing, developing and optimizing the entire software stack that supports their execution on both physical and virtualized infrastructures.

Our research directions include

Approach

Our research work on Big Data analytics platforms is driven by accurate observation of product and client workloads and uses cases. We perform theoretical analysis and benchmarks of state-of-the art solutions to understand their ability to answer the specific needs of emerging relevant workloads, we identify their shortcomings, and we design new solutions and technologies that are able to fill their gaps.

Prototyping and performance evaluation on real large scale deployments has a central role in our research as we aim at providing immedaite value for the development of next-generation Big Data processing systems.

Andrea Reale Kostas Katrinis Mustafa Rafique Bogdan Nicolae Srikumar Venugopal

Resource usage monitoring of an Hadoop deployment on Power8 using the Ganglia Monitoring System.

ganglia monitoring

Overview of a sample Big Data streams processing applications on IBM InfoSphere Streams® Studio9462.

 Big Data Streams