Systems Challenges in Graph Analytics
Explore the complexities of building large-scale, in-memory graph analytics systems in this lecture by Tim Harris from Oracle Laboratories. Delve into the core challenges of graph processing, including searching through billions of records, ranking web pages, and identifying potential connections in social networks. Examine how different graph structures, such as planar road networks and low-diameter social networks, impact algorithm performance and machine resource allocation. Gain insights into the scalability issues faced when operating on SPARC machines with over 4000 threads. Learn about the importance of selecting appropriate algorithms for various graph types and the decision-making process for resource allocation in large-scale systems.