Abstract

Cloud-native data systems today need to address a number of pressing challenges: from handling massive amounts of data using different cloud form factors, to efficiently managing a pool of dis-aggregated hardware resources. This is a non-trivial undertaking, especially in the post-Moore era of increased hardware heterogeneity that opens the potential to offload compute closer to data (e.g., storage) or push operations down to where the data moves (e.g., on the network cards or switches). While this opens a number of exciting opportunities, it is not clear on what the best way forward is to absorb the increased complexity.

In my talk, I argue that addressing such a range of challenges, requires an effort that is beyond what can be typically done within a single layer of the system stack. My proposal calls for a holistic approach that opens up the data system interfaces, embraces the synergy with compilers, and revisits the underlying system support for modern data workloads in virtualized environments.

Bio: Jana Giceva conducts research in the areas of data management and computer systems. Her research interests are in systems support for data science and big data to enable efficient use of modern and future hardware. Prof. Giceva’s research spans across multiple system sub-fields: data processing layer, operating systems, and hardware accelerators for data processing.

Prof. Giceva got her Ph.D. in Computer Science from ETH Zurich in 2017. From 2017 to 2019 she was a Lecturer (Assistant Professor) in the Department of Computing at Imperial College London. She also held roles at Microsoft Research and Oracle Labs during her doctoral studies. She has been a professor for Database Systems at TUM since 2020.