Towards One-Size-Fits-All Data Systems

There is a proliferation of software systems for storing and managing large collections of data. While each of these systems are targeted at different types of data or workloads (e.g. transactions, analytics, graph processing), end users increasingly see a variety of data and workloads in their applications (e.g. transactions and analytics, relational and graph processing). As a result of these heterogenous application requirements, end users have to manage and integrate several types of data management systems, which is tedious, expensive, and counter-productive. So rather than making the world of data management easier, current systems often make the life of the end users much harder.

In this talk, I will describe building scalable data processing systems that ease the pain of supporting multiple types of data and workloads for the end users, i.e., they are easy to deploy and maintain, and they have comparable performance as specialized systems. I will show how modern data management requirements can be supported by providing data storage that is flexible, adaptive, and robust to a variety of query workloads, both in traditional relational databases as well as in new large-scale data flow systems.
Alekh Jindal is a postdoctoral researcher at MIT, working with Samuel Madden on topics in data-intensive systems and database analytics. He received his PhD from Saarland University, where he worked with Jens Dittrich. His PhD research was supported by "Strategic Innovation Fund" from MMCI Cluster of Excellence, Germany and "Validation of Innovation Potential" grant by German Ministry of Education and Science. His research has been published at database venues such as VLDB, SIGMOD, and CIDR, and he received VLDB 2014's Best Paper Award and CIDR 2011's Best Outrageous Ideas and Vision Paper Award.