Bridging the Gap between Relational OLTP and Graph-based OLAP

Explore a groundbreaking conference talk from USENIX ATC '23 that addresses the challenge of performing dynamic graph analytical processing (GAP) tasks on relational OLTP datasets in real-time. Discover GART, an innovative in-memory system that extends hybrid transactional/analytical processing (HTAP) systems to support GAP, creating a hybrid transactional and graph analytical processing (HTGAP) solution. Learn about GART's unique features, including transparent data model conversion through graph extraction interfaces and efficient dynamic graph storage optimized for HTGAP workloads. Understand the key components of GART's architecture, such as the mutable compressed sparse row (CSR) representation, coarse-grained multi-version concurrency control (MVCC) scheme, and flexible property storage. Gain insights into GART's performance advantages, which outperform existing solutions in terms of freshness and efficiency, and surpass state-of-the-art dynamic graph storage systems like LiveGraph by up to 4.4× for GAP workloads on the LDBC SNB dataset.