While more and more applications are distributed in that they execute across multiple devices, software engineers are still provided programming abstractions geared at centralized setups. Reconciling and combining the two perspectives of (a) distributed systems and (b) programming languages and models is challenging yet supports advances in a variety of modern-day application scenarios.

In this talk I will attempt to validate this conjecture underlying my research through several examples in the context of big data analytics. Examples where we achieved significant improvements by slightly increasing distribution awareness include geo-distributed big data analytics (i.e., analytics across multiple datacenters), and privacy-preserving big data analytics. 

Patrick Eugster is a Professor of Computer Science at TU Darmstadt, where he leads the Distributed Systems Programming group. He is also an Associate Professor of CS at Purdue University, on leave. He is interested in distributed systems and programming languages, and in particular in the intersection between the two areas. Particular topics of interest currently include cloud computing, big data analytics, and sensor networks. Patrick holds M.S. and Ph.D. degrees from EPFL. He has been recognized for his research by an NSF CAREER award (2007), an ERC Consolidator grant (2013), and induction into the DARPA Computer Science Study Group (2011) and Purdue University's Inventor Hall of Fame (2015). Besides by various funding agencies including NSF, DARPA, DFG, BMBF, and ERC, his research has been supported by several companies such as Google, Amazon, NetApp, Cisco, HP, and Northrop Grumman.