BARC Talk by Niv Dayan

Tuesday, 21 September 2022,  our new colleague Niv Dayan, Postdoc at BARC and newest addition to the AC section at DIKU, will give a talk about his past and current research.

Title:
Scaling Write-Intensive Storage Engines

Abstract:
Our society is generating data at an exponential rate. Storage engines, such as Google’s Bigtable, Amazon’s DynamoDB, and Facebook’s RocksDB, are the software components that maintain this data and facilitate the extraction of information from it. But exponential data growth exacts a toll on these engines: as the data grows, performance deteriorates. In turn, the applications that run on top must spend disproportionately more time, energy and hardware to, say, complete a transaction on a blockchain, train a deep learning model, or upload a photo to the cloud.

This talk will rethink the architecture of modern storage engines and their underlying data structures to confront the reality of exponential data growth. We will explore new designs that are more scalable and how to generalize them into a universal, tunable data structure that can optimize for diverse applications. 

Bio:
Niv Dayan is a postdoc at DIKU in the Algorithms and Complexity section. He is working on data structure design and analysis for database systems. He spent the last three years as a research scientist at Pliops, a startup based in Tel Aviv. Before that he was a research associate at Harvard, and he completed his PhD at the IT University of Copenhagen. Niv is starting a position as an assistant professor at the University of Toronto in January 2023.