The pressure on the existing software infrastructure continues to increase. On the one hand, there is the exponential growth in data volumes and the need to make data centers more efficient. On the other hand, the hardware landscape has embarked on a major push towards specialization. Both have major implications on software development.

Jana Giceva, Professor of Database Systems at TUM School of Computation, Information and Technology, tackles these challenges in her project “Future-Proof Data Systems” with a holistic approach that establishes the foundations of a new software infrastructure. She proposes to use a memory-centric approach that enables data-intensive jobs to be executed more efficiently on modern hardware and supports reasoning about the different trade-offs regarding performance, cost, and resource efficiency. Furthermore, to facilitate the development of future-proof data systems, she proposes replacing traditional software monoliths with a novel framework for optimization and compilation and a runtime system that is amenable to the evolving nature of modern hardware.