Oasis: An Out-of-core Approximate Graph System via All-Distances Sketches

This conference talk presents Oasis, an Out-of-core Approximate Graph System that leverages All-Distances Sketches (ADS) for efficient large-scale graph processing. Learn how researchers from The Chinese University of Hong Kong and The University of Texas at Dallas developed a solution to the memory constraints of ADS, which typically requires space multiple times larger than the graph data itself. Discover how Oasis creates a holistic framework that enables both efficient ADS construction and estimation by effectively utilizing storage resources. The system allows users to adjust memory usage based on available resources while providing a user-friendly interface that minimizes the impact of slow storage I/O. The presentation demonstrates how Oasis achieves exceptional execution time and low memory usage with only a slight decrease in accuracy, making ADS techniques practical for processing the exponentially growing real-world graphs of today.