SmartGraph: A Framework for Graph Processing in Computational Storage

Graph processing plays a pivotal role in numerous large-scale applications, including social and transportation networks. One of the primary challenges in handling large-scale graph data is its tendency to surpass DRAM capacities. Conventional methods focus on minimizing I/O latency by decreasing disk I/O requests via predictive value calculations. However, these techniques often struggle with inefficient partitioning strategies that elevate DRAM needs, underutilized predictive calculations, and incur considerable synchronization overheads. In our research, we introduce and assess SmartGraph, a new graph partitioning and processing framework that is optimized for both CPU-driven systems and near-storage processing units, such as SmartSSDs. SmartGraph is designed to enhance data-flow within and between processing iterations, drastically reducing the execution latency of graph algorithms and removing synchronization overheads. This framework is especially advantageous in cloud environments, where scalability and efficient data management are paramount. Our experimental findings demonstrate that SmartGraph achieves an average improvement of 1.27x across four graph applications compared to the state-of-the-art framework, LUMOS, when tested on SmartSSDs using datasets like Friendster, LiveJournal, and Twitter. Our empirical analysis underscores the benefits of integrating SmartSSD technology with cloud-based graph processing to boost performance.