July 19, 2016
Talk by Prof. Rui Oliveira, University of Minho, Portugal.
Title: Big Data Management for IoT
Abstract: The proliferation of connected devices, together with the decreasing cost of storage, led to a huge growth on data being generated and collected every day, which can be leveraged by different organizations. However, storing and processing these huge amounts of data in a scalable and cost-effective way poses new challenges. This promoted the emergence of NoSQL database systems and processing solutions based on the MapReduce programming model as alternatives to the traditional Relational Database Management Systems (RDBMS) for large scale data processing. Those solutions, however, trade scalability for programmability, i.e. they sacrifice the standard SQL interface and ACID properties, and they are often specialized for certain workloads (transactional or analytical), requiring the use of different solutions for each workload. The use of multiple data stores has already become a reality and applications often have to handle data from multiple sources, which lack of a common API or programming model.
The CumuloNimbo platform is an alternative solution built on top of NoSQL systems, but that provides the benefits of traditional RDBMS (SQL language, ACID properties), thus maintaining compatibility with existing application stacks, and contributing to improve developers' productivity. First, CumuloNimbo achieves its scalability by decoupling and parallelizing components, and optimizing communications through asynchronous messaging and batching. Second, it offers polyglot capabilities, integrating different data stores under a common query language frameworks, and it will soon enable the aggregation of data arriving from streaming sources with the same API. Finally, it was designed to scale with both transactional and analytical workloads, enabling the use of real-time data on analytics tasks, and eliminating the costs of ETLs and multiple systems.
Bio-sketch: Rui Oliveira graduated in Electrotechnic and Computers Engineering, by the University of Porto, has a Master in Computer Science by the University of Minho and a PhD in Computer Science by The École Polytechnique Fédérale de Lausanne. He is Associate Professor at the Informatics Department of University of Minho and member of the administration board of INESC TEC.
His research interests are on large scale distributed systems, in particular on fault-tolerant agreement protocols, epidemic communication and on exascale data management. Application of his group research has been on scalable distributed database systems and on dependable cloud data storage and processing.
He has published regularly in conferences such as IEEE Dependable Systems and Networks, IEEE Symposium on Reliable Distributed Systems, IEEE International Parallel & Distributed Processing Symposium, IFIP International Conference on Distributed Applications and Interoperable Systems as well as in the IEEE Transactions on Computers. He serves on the steering committees of IEEE SRDS, IFIP/ACM/USENIX Middleware and IFIP DAIS. He is vice-chair of IFIP WG 6.1